WO2024055796A1 - 一种缺陷检测***、方法、装置、电子设备及存储介质 - Google Patents
一种缺陷检测***、方法、装置、电子设备及存储介质 Download PDFInfo
- Publication number
- WO2024055796A1 WO2024055796A1 PCT/CN2023/113105 CN2023113105W WO2024055796A1 WO 2024055796 A1 WO2024055796 A1 WO 2024055796A1 CN 2023113105 W CN2023113105 W CN 2023113105W WO 2024055796 A1 WO2024055796 A1 WO 2024055796A1
- Authority
- WO
- WIPO (PCT)
- Prior art keywords
- image
- product object
- camera
- defect detection
- control device
- Prior art date
Links
- 238000001514 detection method Methods 0.000 title claims abstract description 363
- 230000007547 defect Effects 0.000 title claims abstract description 347
- 238000000034 method Methods 0.000 title claims abstract description 53
- 238000003860 storage Methods 0.000 title claims abstract description 16
- 238000012545 processing Methods 0.000 claims abstract description 59
- 238000013459 approach Methods 0.000 claims abstract description 37
- 230000004044 response Effects 0.000 claims abstract description 7
- 238000007689 inspection Methods 0.000 claims description 46
- 238000010187 selection method Methods 0.000 claims description 46
- 230000003287 optical effect Effects 0.000 claims description 39
- 230000008569 process Effects 0.000 claims description 22
- 238000004590 computer program Methods 0.000 claims description 21
- 238000004891 communication Methods 0.000 claims description 19
- 230000008030 elimination Effects 0.000 claims description 18
- 238000003379 elimination reaction Methods 0.000 claims description 18
- 230000006698 induction Effects 0.000 claims description 16
- 238000012546 transfer Methods 0.000 claims description 4
- 238000012360 testing method Methods 0.000 claims description 3
- 229920000915 polyvinyl chloride Polymers 0.000 description 30
- 239000004800 polyvinyl chloride Substances 0.000 description 30
- 230000002950 deficient Effects 0.000 description 21
- 238000005516 engineering process Methods 0.000 description 21
- 238000004519 manufacturing process Methods 0.000 description 21
- 238000010586 diagram Methods 0.000 description 11
- 230000000007 visual effect Effects 0.000 description 11
- 230000006870 function Effects 0.000 description 10
- 238000011179 visual inspection Methods 0.000 description 9
- 150000002825 nitriles Chemical class 0.000 description 7
- 238000013135 deep learning Methods 0.000 description 6
- 238000013461 design Methods 0.000 description 6
- 230000007246 mechanism Effects 0.000 description 5
- 230000009471 action Effects 0.000 description 3
- 238000004458 analytical method Methods 0.000 description 3
- 238000003384 imaging method Methods 0.000 description 3
- 230000001960 triggered effect Effects 0.000 description 3
- 230000008901 benefit Effects 0.000 description 2
- 238000013527 convolutional neural network Methods 0.000 description 2
- 230000005674 electromagnetic induction Effects 0.000 description 2
- 238000010191 image analysis Methods 0.000 description 2
- 238000012423 maintenance Methods 0.000 description 2
- 239000000463 material Substances 0.000 description 2
- 239000002184 metal Substances 0.000 description 2
- 230000008447 perception Effects 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000005540 biological transmission Effects 0.000 description 1
- 235000014121 butter Nutrition 0.000 description 1
- 230000008878 coupling Effects 0.000 description 1
- 238000010168 coupling process Methods 0.000 description 1
- 238000005859 coupling reaction Methods 0.000 description 1
- 238000013500 data storage Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000018109 developmental process Effects 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 230000005670 electromagnetic radiation Effects 0.000 description 1
- 235000013305 food Nutrition 0.000 description 1
- 238000003709 image segmentation Methods 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 238000009776 industrial production Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 239000013307 optical fiber Substances 0.000 description 1
- 238000012634 optical imaging Methods 0.000 description 1
- 230000002093 peripheral effect Effects 0.000 description 1
- 238000004886 process control Methods 0.000 description 1
- 238000003672 processing method Methods 0.000 description 1
- 238000004451 qualitative analysis Methods 0.000 description 1
- 230000008439 repair process Effects 0.000 description 1
- 230000033764 rhythmic process Effects 0.000 description 1
- 230000011218 segmentation Effects 0.000 description 1
- 238000000926 separation method Methods 0.000 description 1
- 239000007787 solid Substances 0.000 description 1
- 238000007619 statistical method Methods 0.000 description 1
- 238000012549 training Methods 0.000 description 1
- 210000000707 wrist Anatomy 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/0002—Inspection of images, e.g. flaw detection
- G06T7/0004—Industrial image inspection
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20081—Training; Learning
Definitions
- Embodiments of the present application relate to the field of machine vision technology, and in particular to a defect detection system, method, device, electronic equipment and storage medium.
- Disposable gloves such as PVC (polyvinylchloride) gloves and nitrile gloves are in great demand in the medical, industrial, food and other industries. While meeting market demand, the demand for PVC gloves, nitrile gloves and other disposable gloves is increasing. Quality issues are also very important for production safety in various industries. Therefore, taking PVC gloves as an example, while manufacturers of PVC gloves are focusing on studying how to produce PVC gloves efficiently, they also pay more attention to the quality inspection of the produced PVC gloves. In order to achieve high-quality and efficient production activities, manufacturers are choosing to use machine vision inspection solutions to minimize downtime and ensure that safe, high-quality products are always provided.
- PVC polyvinylchloride
- single-sided visual inspection is used to detect defects in PVC gloves.
- PVC gloves that naturally sag on the assembly line will pass through the front side of the background board used to separate multiple rows of PVC gloves driven by the hinge. , enters the visual detection area when passing in front of the background board; the image acquisition mechanism collects images of one side (front or back) of a row of PVC gloves entering the visual detection area, and reports the collected glove images to the control device, thereby controlling the device Detect glove defects based on glove images.
- the purpose of the embodiments of the present application is to provide a defect detection system, method, device, electronic device and storage medium, so as to improve the accuracy of detection results of product object defects.
- the specific technical solutions are as follows:
- an embodiment of the present application discloses a defect detection system, which includes: a control device, a photodetection device and a proximity sensor; wherein, the above photodetection device is provided with a first camera and a third camera arranged in a staggered manner.
- the above-mentioned first camera is a camera that takes pictures of one side of each product object to be inspected passing through the assembly line
- the above-mentioned second camera is a camera that takes pictures of the other side of each product object
- the above-mentioned proximity sensor is arranged on the above-mentioned assembly line, and is configured to send a sensing signal to the above-mentioned control device whenever a product object approaches;
- the above-mentioned control device is configured to encode the currently approaching product object whenever the above-mentioned induction signal is received, obtain the encoding information of the currently approaching product object, and send a photographing trigger signal to the above-mentioned light detection device;
- the above-mentioned photodetection device is configured to respond to the received photographing trigger signal, control the above-mentioned first camera and the second camera to take pictures, obtain the first image and the second image, and report the above-mentioned first image and the second image to the above-mentioned control device;
- the above-mentioned control device is also configured to receive each first image and each second image; and, for each coded information, select the first image and the second image containing the product object indicated by the coded information according to a predetermined image selection method. image, and perform product object defect detection processing on the selected first image and second image, and determine the defect detection result of the product object indicated by the encoding information based on the results obtained by the above product object defect detection processing;
- the image selection method is a selection method set based on the first position difference between the proximity sensor and the first camera and the second position difference between the proximity sensor and the second camera.
- embodiments of the present application disclose a defect detection method based on a defect detection system, which is applied to a control device; the above method includes:
- the currently approaching product object is encoded to obtain the encoding information of the currently approaching product object, and a photographing trigger signal is sent to the above-mentioned light detection device, so that the light detection device responds to the reception
- the above-mentioned first camera and the second camera are controlled to take photos, obtain the first image and the second image, and report the above-mentioned first image and the second image to the above-mentioned control device; wherein, the above-mentioned proximity sensor whenever the product When an object approaches, it sends a sensing signal to the control device;
- For each piece of coded information select a first image and a second image containing the product object indicated by the coded information according to a predetermined image selection method, and perform product object defect detection processing on the selected first image and second image. , based on the results obtained from the above product object defect detection processing, determine the defect detection result of the product object indicated by the encoding information;
- the image selection method is a selection method set based on the first position difference between the proximity sensor and the first camera and the second position difference between the proximity sensor and the second camera.
- a defect detection device based on a defect detection system which is applied to a control device; the above defect detection device includes:
- the sending module is configured to encode the currently approaching product object every time it receives the induction signal sent by the proximity sensor, obtain the encoding information of the currently approaching product object, and send a photographing trigger signal to the above-mentioned light detection device so that the light
- the detection device responds to the received photo-taking trigger signal, controls the first camera and the second camera to take photos, obtains the first image and the second image, and reports the first image and the second image to the above-mentioned control device; wherein, the above-mentioned
- the proximity sensor sends a sensing signal to the control device whenever the product object approaches;
- a receiving module configured to receive each first image and each second image
- the detection module is configured to select, for each coded information, a first image and a second image containing the product object indicated by the coded information according to a predetermined image selection method, and conduct a test on the selected first image and second image.
- the product object defect detection process determines the defect detection result of the product object indicated by the encoding information based on the results obtained by the above-mentioned product object defect detection process; wherein the above-mentioned image selection method is based on the above-mentioned proximity sensor and the first camera.
- the selection method is set by the first position difference and the second position difference between the proximity sensor and the second camera.
- the embodiment of the present application also provides an electronic device, including a processor, a communication interface, a memory, and a communication bus, wherein the processor, the communication interface, and the memory complete communication with each other through the communication bus; the memory is used to store the computer.
- Embodiments of the present application also provide a computer-readable storage medium.
- a computer program is stored in the computer-readable storage medium.
- the steps of the defect detection method are implemented.
- Embodiments of the present application also provide a computer program product containing instructions that, when run on a computer, cause the computer to execute the steps of the above defect detection method.
- the embodiment itself also provides a computer program containing instructions.
- the above computer program When the above computer program is run on a computer, it causes the computer to execute the steps of the above defect detection method.
- the embodiment of the present application provides a defect detection system.
- the system includes a control device, a light detection device and a proximity sensor. Whenever a product object approaches, the proximity sensor sends a sensing signal to the control device; the control device sends a sensing signal to the control device every time.
- the control device sends a sensing signal to the control device every time.
- a sensing signal is received, the currently approaching product object will be encoded to obtain the encoded information, and then a photographing trigger signal will be sent to the photodetection device; the photodetection device will respond to the trigger signal and control the first camera and the third camera.
- the two cameras take pictures to obtain the first image and the second image; the control device will obtain each first image and each second image, and for each coded information, select the third image corresponding to the product object in a predetermined image selection method. An image and a second image, the control device then performs product object defect detection processing on the image.
- the image of the product object on the assembly line acquired by the optical inspection device is an image including multiple sides of the product object
- the control device can determine the information about each product object based on a predetermined image selection method.
- the corresponding images that is, the first image and the second image containing the product object indicated by any encoding information, and then perform product object defect detection processing on the image of each product object, based on the results obtained by the product object defect detection processing,
- the defect detection results on both sides of the corresponding product object can be determined.
- the defect detection system of this solution is based on multi-sided visual defect detection and correlation judgment of the corresponding defect detection results of each side to determine the defect detection of the product object.
- the optical inspection device is provided with a first camera and a second camera arranged in a staggered manner to collect images from multiple sides of the product object. It can be seen that this kind of optical inspection device can achieve simultaneous multi-faceted image collection of product objects through a compact design space on the premise of ensuring full online inspection. Compared with the traditional visual inspection system that detects each side individually, it is greatly improved. improve detection efficiency.
- the encoding of product objects is implemented based on proximity sensors, which can provide implementation conditions for the correlation of multi-faceted detection of the same product object.
- Figure 1 is a schematic structural diagram of a defect detection system provided by an embodiment of the present application.
- Figure 2 is a schematic structural diagram of another defect detection system provided by an embodiment of the present application.
- Figure 3(a) is a schematic diagram of a photodetection device in a defect detection system provided by an embodiment of the present application
- Figure 3(b), Figure 3(c), Figure 3(d) and Figure 3(e) are schematic diagrams of various views of the optical inspection device in a defect detection system provided by an embodiment of the present application;
- Figure 4 is a schematic structural diagram of another defect detection system provided by an embodiment of the present application.
- Figure 5 is a schematic structural diagram of another defect detection system provided by an embodiment of the present application.
- Figure 6 is an overall front view of the equipment of a defect detection system provided by an embodiment of the present application.
- Figure 7 is an overall top view of the equipment of a defect detection system provided by an embodiment of the present application.
- Figure 8 is a schematic flow chart of a defect detection method applied to a control device provided by an embodiment of the present application.
- Figure 9 is a schematic flow chart of a defect detection process provided by an embodiment of the present application.
- Figure 10 is a schematic diagram of a defect detection device provided by an embodiment of the present application.
- FIG. 11 is a schematic structural diagram of an electronic device provided by an embodiment of the present application.
- Machine vision online inspection system is an inspection equipment integrated into the production process. It is an industrial application system that integrates optical, mechanical, electronic, computing, software and other technologies. It detects the spatiotemporal pattern of electromagnetic radiation. For detection and perception, one or more target object images can be automatically acquired, the various characteristic quantities of the acquired images can be processed and analyzed, and qualitative analysis and quantitative interpretation can be made based on the analysis results to obtain the understanding of the relevant target objects and make decisions. Correspondence decision-making is a process control system that makes decisions and performs practical actions.
- Industrial camera A key component in the machine vision system, its most essential function is to convert optical signals into ordered electrical signals.
- Proximity sensor It is a sensor that replaces contact detection methods such as limit switches and detects without contacting the detection object.
- Proximity sensors can detect movement information and presence information of objects and convert them into electrical signals. Detection methods that convert into electrical signals include methods that utilize eddy currents generated in the metal body of the detection target due to electromagnetic induction, methods that capture changes in the capacity of electrical signals caused by the approach of a metal body, and methods that use sharp stones and guide switches. Way.
- Optical inspection and rejection machine equipment a visual system device that uses optical imaging, software, electrical and other technologies to integrate hand mold and glove image collection, defect detection and rejection mechanisms.
- Deep learning image segmentation Based on the deep learning convolutional neural network model, the technology and process of dividing the image into several specific areas with unique properties and proposing targets of interest. It is a key step from image processing to image analysis.
- Deep learning target detection It is based on the deep learning convolutional neural network model to identify the category of the target in the image, or to predict the location of the object, or to determine their location and category for multiple targets.
- the human eye is used to detect the surface defects of the PVC gloves, and then the gloves with surface defects are peeled off from the hand mold manually; in this way, not only more work is required
- the manpower and labor intensity are very high, and the detection accuracy is also low.
- due to cost constraints it is impossible to detect all the gloves produced, and there is a certain probability of omissions.
- single-sided visual inspection is used to detect defects in PVC gloves.
- PVC gloves that naturally sag on the assembly line will pass through the front side of the background board used to separate multiple rows of PVC gloves driven by the hinge. , enters the visual detection area when passing in front of the background board; the image acquisition mechanism collects images of one side (front or back) of a row of PVC gloves entering the visual detection area, and reports the collected glove images to the control device, thereby controlling the device Detect glove defects based on glove images.
- embodiments of the present application provide a defect detection system, method, device, electronic device and storage medium.
- the present application provides a defect detection system, which includes a control device, a light detection device and a proximity sensor; wherein the light detection device is provided with a first camera and a second camera arranged in a staggered manner, and the third camera One camera is a camera that takes pictures of one side of each product object to be inspected passing through the assembly line, and the second camera is a camera that takes pictures of the other side of each product object;
- the above system includes:
- the proximity sensor is arranged on the assembly line and is used to send a sensing signal to the control device whenever a product object approaches;
- the control device is configured to encode the currently approaching product object whenever the induction signal is received, obtain the encoding information of the currently approaching product object, and send a photographing trigger signal to the light detection device;
- the photodetection device is used to control the first camera and the second camera to take pictures in response to the received photographing trigger signal, obtain the first image and the second image, and report the first image and the second image. to the control device;
- the control device is also configured to receive each first image and each second image; and, for each coded information, select the first image and the third image containing the product object indicated by the coded information according to a predetermined image selection method. two images, and perform product object defect detection processing on the selected first image and second image to obtain the defect detection result of the product object indicated by the encoding information;
- the image selection method is a selection method set based on a first position difference between the proximity sensor and the first camera and a second position difference between the proximity sensor and the second camera.
- control device may be an independent control device that communicates with the light detection device and the proximity sensor.
- control device can realize the control function of the control device.
- control device can be a computer equipped with visual inspection software; of course, the functions of the control device can also be realized by at least two devices cooperating with each other. That is to say, the device form of the control device can be at least two devices. The combination of equipment is also reasonable.
- the visual inspection software can encode the induction signal sent by the proximity sensor, send out the photo trigger signal, receive the image sent by the light inspection device, select the image including the product object, and perform product object defect detection processing on the image, and make Decision accordingly.
- the camera provided in the optical detection device may be an industrial camera.
- the optical detection device in this application can rely on the essential function of an industrial camera: convert optical signals into orderly electrical signals, so that after receiving the photo trigger signal, it can perform a photo operation to obtain an image, and then report the image to the control device.
- Proximity sensors are sensors that replace contact detection methods such as limit switches and are designed to detect without touching the detection object.
- the product object can be any product on the assembly line that requires image detection to obtain defect detection results.
- the product object can be disposable gloves such as PVC gloves or nitrile gloves, but of course it is not limited to this.
- control device optical inspection device and proximity sensor in the embodiment of the present application can jointly form a complete machine vision online inspection system to implement defect detection processing of products in the production process, and based on defects Decisions will be made based on the test results.
- the image of the product object on the assembly line acquired by the optical inspection device is an image including multiple sides of the product object
- the control device can determine the information about each product object based on a predetermined image selection method.
- the corresponding images that is, the first image and the second image containing the product object indicated by any encoding information, and then perform product object defect detection processing on the image of each product object, based on the results obtained by the product object defect detection processing,
- the defect detection results on both sides of the corresponding product object can be determined.
- the defect detection system of this solution is based on multi-sided visual defect detection and correlation judgment of the corresponding defect detection results of each side to determine the defect detection of the product object.
- the optical detection device is provided with a first camera and a second camera arranged in a staggered manner to collect images from multiple sides of the product object. It can be seen that this kind of optical inspection device can realize multi-faceted image collection of product objects at the same time through a compact design space on the premise of ensuring full online inspection. Compared with the traditional visual inspection system that detects each side individually, it is greatly improved. improve detection efficiency.
- the encoding of product objects is implemented based on proximity sensors, which can provide implementation conditions for the correlation of multi-faceted detection of the same product object.
- FIG. 1 is a schematic structural diagram of a defect detection system provided by an embodiment of the present application.
- the system includes: a proximity sensor 110, a control device 120, and a photodetection device 130.
- the proximity sensor 110 is connected to the control device 120, and the control device 120 Connected to the photodetection device 130, the photodetection device 130 is provided with a first camera 1301 and a second camera 1302 arranged in a staggered manner.
- the first camera 1301 is used to inspect one side of each product object to be inspected passing through the assembly line.
- the second camera 1302 is a camera that takes pictures of the other side of each product object. in,
- the proximity sensor 110 is provided on the assembly line, and is configured to send a sensing signal to the control device 120 whenever a product object approaches.
- the control device 120 is configured to encode the currently approaching product object whenever the sensing signal is received, and determine the encoding information of the currently approaching product object based on the results obtained by the product object defect detection process, and sending a photographing trigger signal to the photodetection device 130 .
- the optical inspection device 130 is configured to control the first camera 1301 and the second camera 1302 to take pictures in response to the received photographing trigger signal, obtain the first image and the second image, and report the first image and the second image to the control device 120 .
- the control device 120 is also configured to receive each first image and each second image; and, for each coded information, select the first image and the product object indicated by the coded information according to a predetermined image selection method. the second image, and perform product object defect detection processing on the selected first image and second image to obtain the defect detection result of the product object indicated by the encoding information;
- the image selection method is a selection method set based on the first position difference between the proximity sensor 110 and the first camera 1301 and the second position difference between the proximity sensor 110 and the second camera 1302 .
- the defect detection system is a detection equipment integrated in the production process. After the production process, there are already rows of product objects on the assembly line waiting to enter the detection area of the defect detection system, for example: PVC Take gloves as an example. There are two or four rows of PVC gloves that naturally sag on the assembly line and need to be passed. Then, each row of PVC gloves will enter the detection area of the defect detection system driven by the hinge. Of course, this application does not limit the number of rows of PVC gloves present on the assembly line.
- the optical inspection device can be provided with a first camera and a second camera arranged in a staggered manner for each row of product objects. That is to say, the first camera and the second camera to collect images from multiple sides of each product object in the same row of product objects.
- the light detection device 130 can be placed behind the proximity sensor. That is, any product object first passes the proximity sensor, and then passes through the light detection device 130 after being passed for a period of time. In the image acquisition area, at this time, the product object captured by the camera of the light detection device 130 is different from the product object sensed by the proximity sensor 130 .
- the placement position of the photodetection device 130 can also be a position that satisfies the following conditions: when any product object approaches the proximity sensor 110, it simultaneously enters the image collection area of the photodetection device.
- the product object that is close to the proximity sensor and The product object photographed by the target camera in the light detection device belongs to the same product object; wherein, the target camera is the camera close to the proximity sensor 130 among the first camera and the second camera.
- the solution will be introduced below using the first camera as a camera close to the proximity sensor 130 .
- the proximity sensor 110 since the proximity sensor is provided on the assembly line, whenever a product object approaches, the proximity sensor 110 can sense it and send a sensing signal to the control device 120 , so that the control device 120 performs operations on the product object. Coded operations provide conditions. It can be understood that an alternating magnetic field can be generated inside the proximity sensor 110. When a product object hanging on a hinge of the assembly line approaches as the assembly line runs, the hinge will undergo electromagnetic induction with the alternating magnetic field generated by the vibrator to generate Induced current; when product object passes proximity sensor 110 In the process, the proximity sensor 110 can convert the position information and existence information of the product object into sensing signals without touching the product object, and send the sensing signals to the control device 120 .
- the type of proximity sensor 110 can be selected according to the needs of the production site. This application does not limit the specific type of proximity sensor 110 .
- the control device 120 whenever the proximity sensor 110 sends a sensing signal, the control device 120 can receive the sensing signal and encode the currently approaching product object, thereby encoding each position that passes the proximity sensor 110.
- the product objects are assigned identification information, and the product objects at different positions on the assembly line can be distinguished by encoding the information, ultimately providing implementation conditions for the correlation of multi-faceted detection of the same product object.
- control device 120 can send a photographing trigger signal to the photodetection device 130 to trigger the first camera in the photodetection device. and a second camera for image acquisition. That is, every time the product object approaches, the light detection device 130 will be triggered to collect images.
- the photodetection device may also be called a photodetection machine; whenever the photodetection device 130 receives a photographing trigger signal from the control device 120, it may control the first camera 1301 and the second camera 1302 to perform Take a photo to obtain the first image and the second image, and upload the first image and the second image to the control device 120 . It can be understood that since the first camera 1301 and the second camera 1302 are arranged in a staggered manner, and the first camera 1301 and the second camera 1302 shoot at the same time under the trigger of the same photographing trigger signal, the images obtained cannot correspond to the same product object.
- the light detection device 130 triggers the current photographing.
- the first camera 1301 can be triggered to take a picture of the unobstructed side of the product object, that is, the first image.
- the second camera 1302 can be triggered to take a picture of the unobstructed side of the product object.
- the other side of the product object that is, the second image.
- the other photo trigger signal is the next photo trigger of the current photo trigger signal. Signal.
- the first image may be the front side of the product object being photographed, and correspondingly, the second image may be the back side of the product object being photographed; of course, the first image may also be the back side of the product object being photographed, Correspondingly, the second image may be the front side of the photographed product object.
- the target glove has entered the light inspection device.
- the light inspection device responds to the first trigger signal and controls the first camera and the second camera to take pictures.
- the first camera can take pictures of the front side of the target glove.
- the second camera can capture an image that does not include the target glove; as the assembly line runs, the light detection device responds to the second trigger signal and controls the first camera and the second camera to take pictures.
- the first camera can capture an image that does not include the target glove.
- the second camera can capture the back of the target glove.
- the first position difference between the proximity sensor 110 and the first camera 1301 and the second position difference between the proximity sensor 110 and the second camera 1302 are regular features that can be based on The regular feature associates each product object with the first image and the second image containing the product object.
- the control device 120 can identify each product object that is sequentially close to the proximity sensor 130 on the assembly line through the encoded information, so that the first position difference corresponding to each product object can be selected based on the determined first position difference and the second position difference.
- the image and the second image that is, according to the predetermined image selection method, the first image and the second image containing the product object indicated by each encoding information are selected.
- the proximity sensor, the first camera and the second camera are closely arranged, and there is no gap between the components, that is, there is no product object.
- the first product object, the second product object, and the third product object are close to each other in chronological order.
- Product object of proximity sensor, first product object, second product object and third product The objects are adjacent product objects on the assembly line; then, when the first product object approaches the proximity sensor, the first camera and the second camera capture images that do not include the first product object; when the second product object approaches the proximity sensor, the One camera captures a frontal image of the first product object, and the second camera captures an image that does not include any product object; when the third product object approaches the proximity sensor, the first camera captures a frontal image of the second product object, and the second camera captures a frontal image of the second product object.
- the camera captures the back image of the first product object; at this point, the control device can select the front image and the back image of the first product object based on the position difference and the encoding information.
- a product object can enter the image collection area of the first camera while approaching the proximity sensor.
- the first camera and the second camera capture adjacent product objects.
- the first camera can capture the product object that is approaching the sensor, but the second camera cannot capture it.
- the second camera Only then can the image of the product object that passed the proximity sensor be captured, that is, the image of the product object.
- the first image captured by the first camera and the second image captured by the second camera can be used as the image of the product object. Two images to be associated.
- the control device 120 can perform product object defect detection processing on the images, thereby obtaining the defect detection result of the corresponding product object.
- the control device may use a detection model pre-trained for product object defect detection, and then input the first image and the second image into the detection model to perform product object defect detection.
- the establishment of defect detection models can be achieved with the help of deep learning segmentation and target detection technologies.
- the detection model for product object defect detection can use a solution that combines deep learning with traditional 2D algorithms, adding the type of defects judged by the human eye to the network to learn, and thus training the model.
- the embodiments of this application do not limit the specific defect detection process.
- the control device 120 can obtain the defect detection results of the first image and the second image respectively; when the defect detection results of the first image and the second image are When the results all indicate that the product object is defective, that is, under the logical relationship of AND, the result that the corresponding product object is defective can be obtained, that is, the defect detection result of the product object indicated by the encoding information is defective; or, when the third When at least one of the defect detection results of the first image or the second image indicates that the product object is defective, that is, under the logical relationship of OR, the result that the corresponding product object is defective can be obtained, that is, the product indicated by the encoding information The defect detection result of the object is defective; or, when the defect detection results of the first image and the second image both indicate that the product object is not defective, that
- the specific content of the defect detection result of the first image and the specific content of the defect result of the second image may be text content used to indicate whether there is a defect, such as: "defective” or "no defect”. At this time, if If the defect detection result of an image is "defective", it means that one side of the product object is defective. If the defect detection result of an image is "no defect", it means that one side of the product object has no defects.
- the specific content of the defect detection result of the first image and the specific content of the defect result of the second image may be defect category information, wherein the defect category information may include category information used to characterize the absence of a defect, and the category information when a defect exists.
- the category information of the specific defect category to which the defect belongs at this time, if the defect detection result of an image is the category information of the specific defect category, it means that there is a defect on one side of the product object, and if the defect detection result of an image is the indicator that there is no defect
- the category information indicates that one side of the product object has no defects.
- the specific content of the defect detection result of the first image and the specific content of the defect detection result of the second image are When the content can be defect category information, the defect detection result of the product object indicated by the encoding information can also represent the specific category information of the defect.
- the light detection device 130 can be controlled by a stroboscopic light source, which increases the brightness range of the imaging, greatly improves the quality of the imaging, makes the imaging closer to human eye perception, and is compatible with high-speed production line rhythms, while increasing the service life of the hardware. .
- a limiter can also be provided on the bracket used to transfer product objects in the assembly line.
- the limiter whenever the product object approaches the proximity sensor, the limiter can stop the assembly line smoothly and briefly. At this time, the optical inspection device can control the camera to take pictures and obtain clear images.
- the image of the product object on the assembly line acquired by the optical inspection device is an image including multiple sides of the product object
- the control device can determine the information about each product object based on a predetermined image selection method.
- the corresponding images that is, the first image and the second image containing the product object indicated by any encoding information, and then perform product object defect detection processing on the image of each product object, based on the results obtained by the product object defect detection processing,
- the defect detection results on both sides of the corresponding product object can be determined.
- the defect detection system of this solution is based on multi-sided visual defect detection and correlation judgment of the corresponding defect detection results of each side to determine the defect detection of the product object.
- the optical detection device is provided with a first camera and a second camera arranged in a staggered manner to collect images from multiple sides of the product object. It can be seen that this kind of optical inspection device can realize multi-faceted image collection of product objects at the same time through a compact design space on the premise of ensuring full online inspection. Compared with the traditional visual inspection system that detects each side individually, it is greatly improved. improve detection efficiency.
- the encoding of product objects is implemented based on proximity sensors, which can provide implementation conditions for the correlation of multi-faceted detection of the same product object.
- the optical inspection device 130 A third camera 1303 and a fourth camera 1304 are also provided in a staggered arrangement.
- the third camera 1303 is a camera that takes pictures of one side of each product mold passing through the assembly line.
- the fourth camera 1304 is a camera that takes pictures of each product mold. The other side is a camera that takes pictures.
- the photodetection device 130 is also configured to, in response to the received photo-taking trigger signal, control the third camera 1303 and the fourth camera 1304 to take photos, obtain a third image and a fourth image, and send the photo to the control device 120 Report the third image and the fourth image.
- the control device 120 is further configured to receive each third image and each fourth image; and, for each coded information, determine the third image that matches the shooting time of the selected first image, and the third image that matches the shooting time of the selected first image.
- the fourth image whose shooting time matches the second image, perform product mold defect detection processing on the selected third image and fourth image, and determine the mold detection result based on the results obtained by the product mold defect detection processing,
- the obtained mold inspection results are associated with the defect inspection results of the product object indicated by the encoding information.
- the third camera can be positioned above the first camera, and the fourth camera is positioned above the second camera, so that the first camera can Produce
- the product mold of the passing product object can enter the image collection area of the third camera at the same time.
- the product mold associated with the passing product object can be At the same time, enter the image acquisition area of the fourth camera.
- there can be at least one row of product objects for transfer on the assembly line then there can be at least one row of product molds.
- the first camera and the third camera can be arranged in a staggered manner.
- Two cameras, and for each row of product molds, a third camera and a fourth camera are arranged in a staggered manner. That is to say, the first camera and the second camera image multiple sides of each product object in the same row of product objects.
- the third camera and the fourth camera collect images from multiple sides of each product mold in the same row of product molds.
- the schematic diagram of the optical detection device including the first camera 1301, the second camera 1302, the third camera 1303 and the fourth camera 1304 can be shown in Figure 3(a); wherein, within the dotted box The area is the shooting range of the optical path of the camera lens.
- the first camera 1301, the second camera 1302, the third camera 1303 and the fourth camera 1304 are installed on the main body 1305 of the photodetection device.
- Figure 3(b) - Figure 3(e) shows an engineering schematic diagram of various views of the optical detection device.
- the functions implemented by the third camera 1303 and the fourth camera 1304 in the optical detection device 130 can be basically the same as those of the first camera 1301 and the second camera 1302.
- the difference is that the third camera 1303 and the fourth camera 1304
- the photographed object may be a product mold associated with the product object; accordingly, the control device 120 may determine that the image that matches the shooting time of the selected first image is the third image, and determine that the image that matches the shooting time of the selected second image is the third image.
- the image whose shooting time matches is the fourth image.
- the third image captured by the third camera captured at the same time as the first camera that captured the first image can be selected.
- the third image is an image of a specific hand model associated with a specific glove; similarly, a fourth image including a specific hand model associated with a specific glove can be selected.
- the defect detection processing method is the same as the control device 120 performing defect detection processing on the determined first image and second image. The methods are consistent and will not be described in detail in this embodiment.
- the control device 120 can determine the defect detection result of the product object based on the defect detection results of the first image and the second image; wherein the specific determination method has been As described in the previous embodiment; correspondingly, in the case of three logical relationships of AND, OR, and NOT, the control device 120 can also determine the defect detection results of the product mold based on the defect detection results of the third image and the fourth image. .
- the control device 120 can obtain the defect detection results of the third image and the fourth image respectively; when the defect detection results of the third image and the fourth image are When the results all indicate that the product mold is defective, that is, under the logical relationship of AND, you can get the result that the corresponding product mold is defective, that is, the defect detection result of the product mold indicated by the encoding information is defective; or, when the third When at least one of the defect detection results of the three images or the fourth image indicates that the product mold is defective, that is, under the logical relationship of OR, the result that the corresponding product mold is defective can be obtained, that is, the product indicated by the encoding information
- the defect detection result of the mold is defective; or, when the defect detection results of the third image and the fourth image both indicate that the product mold has no defects, that is, under the logical relationship of not, the result that the corresponding product mold can be obtained is that there is no defect. , that is, the defect detection result of the product mold
- the specific content of the defect detection result of the third image and the specific content of the defect result of the fourth image may be text content used to indicate whether there is a defect, such as: "defective” or "no defect". At this time, if If the defect detection result of an image is "defective", it means that one side of the product mold is defective. If the defect detection result of an image is "no defect", it means that one side of the product mold has no defects.
- the specific content of the defect detection result of the third image and the specific content of the defect result of the fourth image may be defect category information, wherein the defect category information may include category information used to characterize the absence of a defect, and the category information when a defect exists.
- the category information of the specific defect category to which the defect belongs at this time, if the defect detection result of an image is the category information of the specific defect category, it means that there is a defect on one side of the product mold, and if the defect detection result of an image is the indicator that there is no defect
- the category information indicates that there are no defects on one side of the product mold.
- the control device 120 can associate the mold defect detection result with the defect detection result of the product object.
- the defect detection result of the target hand mold shows that the hand mold has crack defects
- the defect detection result of the target glove hanging below the target hand mold shows that the glove has damage and crack defects; then, the system can detect the damaged gloves.
- the crack defect is related to the crack defect in the hand mold.
- the defect detection results of the target hand mold show that the hand mold has dirt defects
- the defect detection results of the target glove hanging below the target hand mold show that the gloves have defects such as oil stains and flow marks; then, the system can Correlate oily gloves and flow mark defects with dirty hand mold defects.
- the defect detection system can also perform hierarchical processing on the defects of the product object.
- the control device detects a defect that seriously affects the quality of the product object, the control device can classify the product mold defect associated with the product object defect. Make announcements and notifications to remind relevant personnel to perform maintenance and replacement of product molds.
- the control device detects that the glove has a damage defect, and the hand mold defect detection result associated with the glove shows that the hand mold has a crack defect; then, the system can send the information that the hand mold has a crack defect to the relevant personnel. , to remind relevant personnel to perform repairs on the hand mold.
- control device can, after detecting product object defects and product mold defects, save the product object defects and product mold defect data, and perform statistical analysis, so that the product object can be analyzed with the support of the data. Defects and product mold defects are predicted.
- the defect detection system can detect defects on the product object and the product mold at the same time, and correlate the defect detection results of the two, which can further provide basic data for the analysis of the causes of defects.
- the control device 120 includes: a main control device 410 and a programmable logic controller (Programmable Logic Controller) PLC. 420;
- a programmable logic controller Programmable Logic Controller
- the PLC 420 is configured to encode the currently approaching product object whenever it receives the sensing signal, and sends the encoding information to the main control device 410, and sends a photographing trigger signal to the light detection device 130.
- the main control device 410 is configured to receive each first image and each second image; and, for each encoding information, select the first image and the product object indicated by the encoding information according to a predetermined image selection method. second image, perform product object defect detection processing on the selected first image and second image, and determine the defect detection result of the product object indicated by the encoding information based on the results obtained by the product object defect detection processing.
- the photodetection device 130 is specifically configured to report the first image and the second image to the main control device 410 .
- control device 120 can be subdivided into the main control device 410 and the editable logic controller PLC 420, the functions are also divided accordingly.
- PLC 420 can receive every induction signal, and each induction signal represents each approaching product object. Therefore, the PLC can encode the currently approaching product object, and then can send the encoding information to the main control device 410 and send a photographing trigger signal to the light detection device 130 .
- the main control device 410 can receive each first image and each second image; and for each coded information, select the first image and the second image containing the product object indicated by the coded information according to a predetermined image selection method, The product object defect detection process is performed on the selected first image and the second image to obtain the defect detection result of the product object indicated by the encoding information.
- the main control device 410 and the PLC 420 are separate devices, and the main control device 410 can be any form of electronic equipment, which is not limited by the embodiment of the present application.
- the control device is composed of the main control device 410 and the PLC 420.
- the functions of the control device can be realized through the cooperation of the main control device 410 and the PLC 420.
- functional decoupling can be achieved, thereby improving the convenience of system maintenance and reducing operating costs.
- control device 120 selects the first image and the second image containing the product object indicated by the encoding information according to a predetermined image selection method for each encoding information. Includes step A1:
- Step A1 for each piece of coded information, determine the first image containing the product object indicated by the coded information based on the order corresponding to the coded information, the order corresponding to each first image, and the first position difference; and, based on The order corresponding to the encoding information, the order corresponding to each second image and the second position difference determine the second image containing the product object indicated by the encoding information;
- the order corresponding to the coded information is the order corresponding to the coded information when the respective coded information is sorted according to the order of generation time;
- the order corresponding to each first image is the order corresponding to the first images when the respective first images are sorted according to the shooting time.
- the order corresponding to each second image is the order corresponding to the second images when the second images are sorted according to the shooting time.
- each coded information can be sorted according to the order of generation time, so as to obtain the order corresponding to each coded information, that is, the arrangement position of each coded information in the queue composed of each coded information; and, can be based on the shooting Sort each first image in time to obtain the order corresponding to each first image, that is, the arrangement position of each first image in the queue composed of each first image; similarly, each second image can be sorted according to the shooting time. The images are sorted to obtain the order corresponding to each second image, that is, the arrangement position of each second image in the queue composed of each second image.
- the encoded information of the same product object, the first image and the second image will have different positions in their respective queues, and There is a regularity. Therefore, for each piece of coded information, based on the order corresponding to the coded information, the order corresponding to each first image, and the first position difference, the first image containing the product object indicated by the coded information can be determined. ; And, based on the order corresponding to the encoding information, the order corresponding to each second image, and the second position difference, determine the second image containing the product object indicated by the encoding information. In this way, the first image and the second image of the same product object can be quickly obtained.
- the order obtained by sorting each coded information according to the order of generation time can be a sequential queue, and the time interval between each coded information in the queue can depend on the running speed of the production site assembly line; accordingly, The time interval between each first image and each second image may be consistent with the time interval between encoded information.
- the order obtained by sorting the coded information can form a coded information queue
- the order obtained by sorting each first image can form a first image queue
- the order obtained by sorting each second image can form a second image queue; these three
- the time interval between elements in the queue can be consistent, and the position of each element in the queue can be known.
- the first camera and the second camera can take pictures; accordingly, there can be one more coded information element about the product object in the coded information queue, and there can be one more coded information element about the product object in the first image queue that can be captured by the first camera.
- the second image queue may have one more image element captured by the second camera.
- the position difference between the proximity sensor, the first camera and the second camera is known, that is, when the number of difference elements between the proximity sensor, the first camera and the second camera is known, it can be determined that the corresponding encoding information is included
- the first image and the second image of the product object are indicated.
- the first position difference may be: when the product object to be photographed enters the shooting field of view of the first camera and there is a product object approaching the proximity sensor, the first position difference may be entered into the shooting field of view of the first camera. There is an interval of N1 product objects between the product object and the product object close to the proximity sensor;
- the second position difference may be: when the product object to be photographed enters the shooting field of view of the second camera and there is a product object approaching the proximity sensor, the product object entering the shooting field of view of the second camera is different from the product approaching the proximity sensor.
- N2 product objects between objects among them, N1 and N2 are different values.
- control device determines the first image containing the product object indicated by the encoding information based on the order corresponding to the encoding information, the order corresponding to each first image, and the first position difference, including:
- the control device determines the second image containing the product object indicated by the encoding information based on the order corresponding to the encoding information, the order corresponding to each second image, and the second position difference, including:
- the individual product objects separated between the product object to be photographed and the new product object can be Numbers are used as units of distance. For example, in the glove production line, there are 5 gloves between the first camera and the proximity sensor. Glove No. 0 enters the shooting field of view of the first camera. At the same time, a new glove approaches the proximity sensor. Then the new glove can be recorded as 6 No. 1 gloves.
- N1 and N2 can be non-zero natural numbers; in particular, N1 can be zero when the proximity sensor is closely connected to the first camera.
- the first image when selecting the first image, the first image may be selected from a queue composed of first images, and the first images in the queue are arranged in the order in which they were photographed; it can be understood that, The first image that is sorted first has a smaller order in the queue. Therefore, select the first image whose corresponding order is not less than the order corresponding to the encoding information and differs by N1+1.
- the selected first image corresponds to The order, the order corresponding to the encoded information, differs by N1+1, and the order corresponding to the first image is not less than the order corresponding to the encoded information, that is, the first camera is selected to capture N1 product objects after the pipeline starts from the proximity sensor.
- image which is the N1+1th image.
- the proximity sensor is 5 product objects away from the first camera.
- the system will select the image with an order of 6 in the first image queue to obtain the first image of the product object.
- the selection rules for the second image are the same as those for the first image, except that the distance N1 between the proximity sensor and the first camera is different from the distance N2 between the proximity sensor and the second camera.
- a second image whose corresponding order is not less than the order corresponding to the encoded information and which differs by N2+1 can be selected. That is to say, the order corresponding to the selected second image and the order corresponding to the encoded information differ by N2+ 1.
- the order corresponding to the second image is not less than the order corresponding to the encoded information, that is, the image captured by the second camera after the pipeline runs N2 product objects starting from the proximity sensor is selected, which is the N2+1th image.
- control device 120 may also be configured to: starting from the first first image captured by the first camera, encode each first image in order according to the shooting time to obtain each first image.
- the encoding value and starting from the first second image captured by the second camera, encode each second image in order according to the shooting time to obtain the encoding value of each second image. That is, starting from the first first image, each first image is encoded in order, and the order of each first image is represented by encoding; starting from the first second image, each second image is encoded in order. Encoding is performed, and the order of each second image is represented by encoding.
- Step B1 The control device 120 selects from each first image a first image whose corresponding order is not less than the order corresponding to the encoded information and differs by N1+1, which may include step B11:
- the B11 from each first image, determine the first image whose encoding value meets the first condition, and obtain the first image whose corresponding order is not less than the order corresponding to the encoding information and has a difference of N1+1; wherein, the The first condition is that the coded value is not less than the order corresponding to the coded information and the difference is N1+1.
- Step B2 The control device 120 selects from each second image a second image whose corresponding order is not less than the order corresponding to the encoding information and differs by N2+1, which may include step B21:
- each second image determines the second image whose encoding value meets the second condition, and obtain a second image whose corresponding order is not less than the order corresponding to the encoding information and has a difference of N2+1; wherein, the The second condition is that the coded value is not less than the order corresponding to the coded information and the difference is N2+1.
- the encoding values of the first image and the second image are values representing the order.
- each first image can be encoded in order according to the shooting time.
- each first image captured by the first camera can be There are images that do not contain product objects. It should be noted that the method of selecting the first image and the second image corresponding to the product object may be consistent with steps B1 and B2.
- the distance between the proximity sensor, the first camera, and the second camera is the same, the running speed of the assembly line remains unchanged, the frequency of each glove approaching the proximity sensor remains unchanged, and the photo frequency of the first camera and the second camera remains unchanged. Remain unchanged; each time a photo is taken, the position of each glove is as shown in Table 1; among them, gloves No. 1, Glove No. 2, and Glove No. 3 are the gloves that approach the proximity sensor in chronological order; captured by the first camera Each first image of can be encoded as image No. 1, image No. 2 and image No. 3 in time sequence; similarly, each second image captured by the second camera can also be encoded as image No. 1, image No. 1, image No. 2 and image No. 3 in time sequence. Image No.
- the control device can select the image captured by the first camera.
- the No. 2 image and the No. 3 image captured by the second camera are used as the front image and back image of the No. 1 glove.
- control device 120 may also be configured to: starting from the next image of the N1+1th first image captured by the first camera, process each first image according to the shooting time. Sequential encoding to obtain the encoding value of each first image; and starting from the next image of the N2+1 second image captured by the second camera, sequential encoding of each second image according to the shooting time to obtain each second The encoded value of the image.
- Step B1 The control device 120 selects from each first image a first image whose corresponding order is not less than the order corresponding to the encoding information and differs by N1+1, which may include step B12:
- Step B2 The control device 120 selects from each second image a second image whose corresponding order is not less than the order corresponding to the encoded information and differs by N2+1, which may include step B22:
- each first image is encoded in order according to the shooting time
- the N2+1th image captured by the second camera is Starting from the next image of the second image
- each second image is encoded in order according to the shooting time.
- each first image and each second image captured by the first camera and the second camera may include Image of the product object. It should be noted that the method of selecting the first image and the second image corresponding to the product object may be consistent with steps B1 and B2.
- the distance between the proximity sensor and the first camera is 1 distance between adjacent gloves; the distance between the proximity sensor and the second camera is 2 distances between adjacent gloves, and the running speed of the assembly line remains unchanged.
- the frequency of each glove approaching the proximity sensor remains unchanged, and the frequency of taking pictures by the first camera and the second camera remains unchanged; when each picture is taken, the position of each glove is shown in Table 2; among them, No. 1 glove, 2 Gloves No. 3, Gloves No. 4, and Gloves No. 5 are gloves that approach the proximity sensor in chronological order; starting from the third image captured by the first camera, each first image captured by the first camera is It is coded into image No. 3, image No. 4 and image No.
- each second image taken by the second camera is in the order of shooting time; similarly, starting from the 4th image taken by the second camera, each second image taken by the second camera is in the order of shooting time.
- image No. 4 image No. 5 and image No. 6; then, when glove No. 3 approaches the proximity sensor, the first camera captures image No. 3 containing the front image of glove No. 2, and the second camera captures image No. 3 containing the front image of glove No. 1.
- the control device can select the image captured by the first camera.
- the image No. 4 and the image No. 6 captured by the second camera are used as the image of glove No. 3.
- control device can perform corresponding coding on the captured images containing product objects or images not containing product objects to obtain the image order.
- the order corresponding to the product object coding information and the corresponding order of each image can be The order and the position difference of each component in the system are used to select the image of the corresponding product object to facilitate the defect detection of the product object, thereby further improving the accuracy of the defect detection results of the product object.
- Figure 5 based on the defect detection system shown in Figure 1, it also includes: a rejection device 510, wherein the position of the rejection device on the assembly line is at After the photodetection device 130;
- the control device 120 is also configured to control the rejection device 510 to remove the product object indicated by the coding information from the assembly line when the defect detection result of the product object indicated by any coding information is detected and meets the rejection conditions.
- the elimination device 510 is configured to eliminate the object represented by the encoding information from the pipeline under the control of the control device 120 .
- the rejecting device can also be called a rejecting machine, and the rejecting device can be used to reject non-compliant product objects.
- the position difference between the proximity sensor 110 and the rejection device 510 is: when a product object enters the operable area of the rejection device and a product object approaches the proximity sensor, enter There is an interval of N3 product objects between the product object in the operable area and the product object close to the proximity sensor.
- the control device controls the elimination device to eliminate the product object indicated by the encoding information from the assembly line, which may include the steps:
- an object removal instruction is sent to the removal device, so that the removal device removes the product object indicated by the encoding information from the assembly line;
- the designated sensing signal is a sensing signal emitted when the product object represented by the encoding information approaches.
- control device 120 when it detects the defect detection result of the product object and meets the elimination conditions, it can send an instruction to eliminate the product object to the elimination device 510; the elimination device 510 can receive the instruction sent by the control device and perform the processing of the third product object.
- N3+1 product objects perform the elimination operation. For example, there are 6 product objects between the control device and the rejection device. The control device detects that product object No. 0 needs to be removed. When product object No. 7 happens to be close to the proximity sensor, product No. 0 enters the operation area of the rejection device. Therefore, the control device can issue a rejection instruction when receiving the sensing signal of the No. 7 product object.
- the elimination condition may be that the defect detection results of the product object include predetermined elimination defects. For example, on the glove production line, when the control device detects that the gloves to be tested have broken fingers, holes, large tears, scratched wrists, black oil stains, butter stains, secondary materials, cracks, flow marks, and remaining materials. When there is a defect, the rejecting device is controlled to reject the glove to be tested from the assembly line.
- the embodiments of this application do not limit specific elimination conditions.
- the rejecting device 510 can be applied in a scenario where the light detection device 130 includes a first camera, a second camera, a third camera and a fourth camera.
- the rejecting device 510 can be connected after the optical inspection device for rejecting specified product objects from the assembly line.
- the rejection device may include an equipment body and a grapple.
- the grapple may be controlled by a controllable robotic arm placed on the equipment body. After the controllable robotic arm receives the rejection instruction sent by the control device, , the target glove can be removed by pulling the glove with the grapple.
- a rejection device is added to the defect detection system, and the control device is equipped with the function of a corresponding rejection device, so that unqualified product objects can be removed in a timely manner on the production line of product objects and the quality of product objects can be improved.
- the defect detection system provided by this embodiment is applied to the glove production line, and may include a proximity sensor (not shown in Figure 6), a limiter 01, a light detection device 130, a rejection device 510 and Cabinet 04 with control device deployed.
- FIG. 7 is a top view of FIG. 6 .
- the light detection device 130 can acquire images of multiple sides of gloves on the assembly line, and the control device can determine the corresponding image of each glove based on a predetermined image selection method, that is, the image containing the glove indicated by any encoding information.
- the first image and the second image and then perform glove defect detection processing on the image of each glove to obtain the multi-sided defect detection results of the gloves;
- the control device can determine the gloves that need to be eliminated based on the glove defect detection results, and send the elimination signal Sent to the rejecting device 510;
- the rejecting device 510 can perform the rejecting action and reject the gloves that need to be removed.
- the photodetection device 130 and the rejection device 510 may be combined into a photodetection rejector device. This achieves the coupling of glove image acquisition and defective glove removal functions, improving the efficiency of defect detection and defect removal.
- the defect detection system provided by the embodiment of the present application is a machine vision online detection system, which is composed of a proximity sensor, a limiter, a light detection device, a rejection device and a control device.
- This system can not only realize timely feedback of detection and analysis results to equipment execution, but also It can also better adapt to different production lines and achieve flexible splitting and linking.
- this defect detection method is applied to the control device of the defect detection system and may include the following steps:
- S803 For each coded information, select the image containing the coded information indicated by the predetermined image selection method. The first image and the second image of the product object, and perform product object defect detection processing on the selected first image and the second image to obtain the defect detection result of the product object indicated by the encoding information;
- the image selection method is a selection method set based on a first position difference between the proximity sensor and the first camera and a second position difference between the proximity sensor and the second camera.
- step S803 which selects the first image and the second image containing the product object indicated by the encoding information according to a predetermined image selection method for each encoding information, may include step A1:
- A1 for each encoding information, determine the first image containing the product object indicated by the encoding information based on the order corresponding to the encoding information, the order corresponding to each first image, and the first position difference; and, based on the The order corresponding to the encoding information, the order corresponding to each second image and the second position difference determine the second image containing the product object indicated by the encoding information;
- the order corresponding to the coded information is the order corresponding to the coded information when the respective coded information is sorted according to the order of generation time;
- the order corresponding to each first image is the order corresponding to the first images when the respective first images are sorted according to the shooting time.
- the order corresponding to each second image is the order corresponding to the second image when the respective second images are sorted according to the shooting time;
- the first position difference is: when the product object to be photographed enters the shooting field of view of the first camera and there is a product object approaching the proximity sensor, the product object entering the shooting field of view of the first camera is different from the product object approaching the proximity sensor.
- N1 product objects between objects There are N1 product objects between objects.
- the second position difference is: when the product object to be photographed enters the shooting field of view of the second camera and there is a product object approaching the proximity sensor, the difference between the product object entering the shooting field of view of the second camera and the object approaching the proximity sensor interval, N2 product objects are spaced; among them, N1 and N2 are different values.
- Step A1 Determining the first image containing the product object indicated by the encoding information based on the order corresponding to the encoding information, the order corresponding to each first image, and the first position difference may include step B1:
- B1 From each first image, select the first image whose corresponding order is not less than the order corresponding to the encoding information and which differs by N1+1.
- Step A1 Determining the second image containing the product object indicated by the encoding information based on the order corresponding to the encoding information, the order corresponding to each second image, and the second position difference may include step B2:
- B2 From each second image, select a second image whose corresponding order is not less than the order corresponding to the encoding information and which differs by N2+1.
- the method further includes: starting from the first first image captured by the first camera, encoding each first image in order according to the shooting time to obtain the encoding of each first image. value; and starting from the first second image captured by the second camera, encode each second image in order according to the shooting time to obtain the encoding value of each second image;
- Step B1 Selecting from each first image a first image whose corresponding order is not less than the order corresponding to the encoding information and differs by N1+1 may include step B11:
- B11 From each first image, determine the first image whose encoding value meets the first condition, and obtain the first image whose corresponding order is not less than the order corresponding to the encoding information and has a difference of N1+1; wherein, the The first condition is having There is a coded value that is not less than the sequence corresponding to the coded information and differs by N1+1.
- Step B2 Selecting from each second image a second image whose corresponding order is not less than the order corresponding to the encoding information and differs by N2+1 may include step B12:
- B12 From each second image, determine the second image whose encoding value meets the second condition, and obtain a second image whose corresponding order is not less than the order corresponding to the encoding information and has a difference of N2+1; wherein, the The second condition is that the coded value is not less than the order corresponding to the coded information and the difference is N2+1.
- the method may further include: starting from the next image of the N1+1th first image captured by the first camera, sequentially processing each first image according to the shooting time. Encoding to obtain the encoding value of each first image; and starting from the next image of the N2+1 second image captured by the second camera, encoding each second image in order according to the shooting time to obtain each second image the coded value;
- Step B1 Selecting from each first image a first image whose corresponding order is not less than the order corresponding to the encoding information and differs by N1+1, may also include step B21:
- Step B2 Selecting from each second image a second image whose corresponding order is not less than the order corresponding to the encoding information and differs by N2+1, may also include step B22:
- the defect detection method applied to the control device of the defect detection system may further include step C1:
- the elimination device When the defect detection result of the product object indicated by any encoding information is detected and the elimination condition is met, the elimination device is controlled to eliminate the product object indicated by the encoding information from the assembly line.
- Step C1 controlling the elimination device to eliminate the product object indicated by the encoding information from the assembly line may also include step C11:
- C11 After receiving the specified induction signal, when the N3+1th induction signal is received, send an object removal instruction to the removal device, so that the removal device removes the product object indicated by the encoding information from the assembly line.
- the designated sensing signal is a sensing signal emitted when the product object represented by the encoding information approaches.
- the defect detection method applied to the control device of the defect detection system may further include the following steps:
- D1 receives each third image and each fourth image.
- D2 For each coded information, determine the third image that matches the shooting time of the selected first image, and the fourth image that matches the shooting time of the selected second image. For the selected third image, The image and the fourth image are subjected to product mold defect detection processing to obtain a mold detection result, and the obtained mold detection result is associated with the defect detection result of the product object indicated by the encoding information.
- the image of the product object on the assembly line acquired by the optical inspection device is an image including multiple sides of the product object
- the control device can determine the information about each product object based on a predetermined image selection method.
- the corresponding image that is, the first image and the second image containing the product object indicated by any encoding information, and then the related Product object defect detection processing is performed on the image of each product object.
- the defect detection results on both sides of the corresponding product object can be determined.
- the defect detection system of this solution is based on multi-sided visual defect detection and correlation judgment of the corresponding defect detection results of each side to determine the defect detection of the product object.
- the optical detection device is provided with a first camera and a second camera arranged in a staggered manner to collect images from multiple sides of the product object. It can be seen that this kind of optical inspection device can realize multi-faceted image collection of product objects at the same time through a compact design space on the premise of ensuring full online inspection. Compared with the traditional visual inspection system that detects each side individually, it is greatly improved. improve detection efficiency.
- the encoding of product objects is implemented based on proximity sensors, which can provide implementation conditions for the correlation of multi-faceted detection of the same product object.
- the defect detection process is introduced below by taking the steps of each component in the assembly line to implement defect detection according to the process. As shown in Figure 9, from the perspective of the flow of each component in the assembly line, the defect detection process can include the following steps:
- the proximity sensor sends a sensing signal to the PLC.
- S902 Whenever the PLC receives a sensing signal, it encodes the currently approaching product object, sends the encoding information to the main control device, and sends a photographing trigger signal to the light detection device.
- steps S903 and S906 are executed simultaneously.
- the light detection device responds to the received photographing trigger signal, controls the first camera and the second camera to take pictures, obtains the first image and the second image, and reports the first image and the second image to the main control device. .
- S904 The main control device receives each first image and each second image.
- the main control device For each coded information, the main control device selects the first image and the second image that contain the product object indicated by the coded information according to the predetermined image selection method, and performs operations on the selected first image and second image.
- the product object defect detection process determines the defect detection result of the product object indicated by the encoding information based on the results obtained by the product object defect detection process.
- step S906 The light detection device responds to the received photo-taking trigger signal, controls the third camera and the fourth camera to take photos, obtains the third image and the fourth image, and reports the third image and the fourth image to the main control device. . After step S906, step S907 is executed.
- S907 The main control device receives each third image and each fourth image.
- the main control device determines the third image that matches the shooting time of the selected first image, and the fourth image that matches the shooting time of the selected second image.
- the third image and the fourth image are subjected to product mold defect detection processing, and the mold detection result is determined based on the results obtained by the product mold defect detection processing.
- Step S909 is executed after both step S905 and step S908.
- the main control device associates the obtained mold detection results with the defect detection results of the product object indicated by the encoding information.
- step S9010 The main control device determines whether the product object indicated by the encoding signal needs to be removed from the assembly line. If yes, execute step S9011; if no, execute step S9013;
- the main control device can determine whether the product object needs to be eliminated according to predetermined elimination rules.
- the main control device sends a removal instruction to the removal device.
- the main control device can send a removal instruction to the removal device;
- the rejecting device receives the rejecting instruction sent by the main control device and executes the rejecting operation.
- the rejecting device can receive the rejecting instruction issued by the main control device, and perform the rejecting operation using the rejecting means predetermined by the rejecting device.
- the main control device does not send the rejection command to the rejection device.
- the image of the product object on the assembly line acquired by the optical inspection device is an image including multiple sides of the product object
- the control device can determine the information about each product object based on a predetermined image selection method.
- the corresponding images that is, the first image and the second image containing the product object indicated by any encoding information, and then perform product object defect detection processing on the image of each product object, based on the results obtained by the product object defect detection processing,
- the defect detection results on both sides of the corresponding product object can be determined.
- the defect detection system of this solution is based on multi-sided visual defect detection and correlation judgment of the corresponding defect detection results of each side to determine the defect detection of the product object.
- the optical detection device is provided with a first camera and a second camera arranged in a staggered manner to collect images from multiple sides of the product object. It can be seen that this kind of optical inspection device can realize multi-faceted image collection of product objects at the same time through a compact design space on the premise of ensuring full online inspection. Compared with the traditional visual inspection system that detects each side individually, it is greatly improved. improve detection efficiency.
- the encoding of product objects is implemented based on proximity sensors, which can provide implementation conditions for the correlation of multi-faceted detection of the same product object.
- the defect detection device is applied to the control device of the defect detection system and may include the following modules:
- the sending module 1010 is configured to encode the currently approaching product object whenever it receives an induction signal sent by the proximity sensor, obtain the encoding information of the currently approaching product object, and send a photographing trigger signal to the light detection device to causing the light detection device to respond to the received photographing trigger signal, control the first camera and the second camera to photograph, obtain the first image and the second image, and report the first image and the second image to the control Device; wherein the proximity sensor sends a sensing signal to the control device whenever the product object approaches;
- the receiving module 1020 is configured to receive each first image and each second image
- the detection module 1030 is configured to select the first image and the second image containing the product object indicated by the encoding information according to a predetermined image selection method for each encoding information, and perform the selected first image and the second image Perform product object defect detection processing to obtain the defect detection result of the product object indicated by the encoding information; where,
- the image selection method is a selection method set based on a first position difference between the proximity sensor and the first camera and a second position difference between the proximity sensor and the second camera.
- the image of the product object on the assembly line acquired by the optical inspection device is an image including multiple sides of the product object
- the control device can determine the information about each product object based on a predetermined image selection method.
- the corresponding images that is, the first image and the second image containing the product object indicated by any encoding information, and then perform product object defect detection processing on the image of each product object, based on the results obtained by the product object defect detection processing,
- the defect detection results on both sides of the corresponding product object can be determined.
- the defect detection system of this solution is based on multi-sided visual defect detection and correlation judgment of the corresponding defect detection results of each side to determine the defect detection of the product object.
- the optical detection device is provided with a first camera and a second camera arranged in a staggered manner to collect images from multiple sides of the product object. It can be seen that this kind of optical inspection device can realize multi-faceted image collection of product objects at the same time through a compact design space on the premise of ensuring full online inspection. Compared with the traditional visual inspection system that detects each side individually, it is greatly improved. improve detection efficiency.
- the encoding of product objects is implemented based on proximity sensors, which can provide implementation conditions for the correlation of multi-faceted detection of the same product object.
- the embodiment of the present application also provides an electronic device, as shown in Figure 11, including a processor 1101, a communication interface 1102, a memory 1103, and a communication bus 1104.
- the processor 1101, the communication interface 1102, and the memory 1103 communicate through the communication bus 1104. complete mutual communication,
- the processor 1101 is used to implement the above defect detection method when executing the program stored in the memory 1103.
- the communication bus mentioned in the above-mentioned electronic equipment can be a Peripheral Component Interconnect (PCI) bus or an Extended Industry Standard Architecture (EISA) bus, etc.
- PCI Peripheral Component Interconnect
- EISA Extended Industry Standard Architecture
- the communication bus can be divided into address bus, data bus, control bus, etc. For ease of presentation, only one thick line is used in the figure, but it does not mean that there is only one bus or one type of bus.
- the communication interface is used for communication between the above-mentioned electronic devices and other devices.
- the memory may include random access memory (Random Access Memory, RAM) or non-volatile memory (Non-Volatile Memory, NVM), such as at least one disk memory.
- RAM Random Access Memory
- NVM Non-Volatile Memory
- the memory may also be at least one storage device located far away from the aforementioned processor.
- the above-mentioned processor can be a general-purpose processor, including a central processing unit (CPU), a network processor (Network Processor, NP), etc.; it can also be a digital signal processor (Digital Signal Processor, DSP), special integrated Circuit (Application Specific Integrated Circuit, ASIC), Field-Programmable Gate Array (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components.
- CPU central processing unit
- NP Network Processor
- DSP Digital Signal Processor
- ASIC Application Specific Integrated Circuit
- FPGA Field-Programmable Gate Array
- a computer-readable storage medium is also provided.
- the computer-readable storage medium A computer program is stored in the storage medium, and when the computer program is executed by the processor, the steps of any of the above defect detection methods are implemented.
- a computer program product containing instructions is also provided, which when run on a computer causes the computer to execute any of the defect detection methods in the above embodiments.
- a computer program containing instructions is also provided.
- the computer program When the computer program is run on a computer, it causes the computer to execute any of the defect detection methods in the above embodiments.
- the computer program product includes one or more computer instructions.
- the computer may be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device.
- the computer instructions may be stored in or transmitted from one computer-readable storage medium to another, e.g., the computer instructions may be transferred from a website, computer, server, or data center Transmission to another website, computer, server or data center by wired (such as coaxial cable, optical fiber, digital subscriber line (DSL)) or wireless (such as infrared, wireless, microwave, etc.) means.
- the computer-readable storage medium may be any available medium that can be accessed by a computer or a data storage device such as a server, data center, etc. that contains one or more available media integrated.
- the available media may be magnetic media (eg, floppy disk, hard disk, tape), optical media (eg, DVD), or other media (eg, Solid State Disk (SSD)), etc.
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Biophysics (AREA)
- Evolutionary Computation (AREA)
- Health & Medical Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- Biomedical Technology (AREA)
- Quality & Reliability (AREA)
- Computational Linguistics (AREA)
- Data Mining & Analysis (AREA)
- Computer Vision & Pattern Recognition (AREA)
- General Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Investigating Materials By The Use Of Optical Means Adapted For Particular Applications (AREA)
Abstract
本申请实施例提供了一种缺陷检测***、方法、装置、电子设备及存储介质,涉及机器视觉技术领域。该***中,接近传感器每当产品对象接近时,向控制装置发送感应信号;控制装置接收到感应信号,针对产品对象进行编码,得到编码信息,向光检装置发送触发信号;光检装置响应触发信号,控制第一相机和第二相机进行拍照,得到第一图像和第二图像;控制装置接收各个第一图像和各个第二图像;针对每一编码信息,按照预定图像选取方式,选取包含编码信息所指示的产品对象的第一图像和第二图像,并对所选取的第一图像以及第二图像进行产品对象缺陷检测处理,确定编码信息指示的产品对象缺陷检测结果。可见,上述***可以提高产品对象缺陷检测结果的准确率。
Description
本申请要求于2022年9月14日提交中国专利局、申请号为202211116759.7发明名称为“一种缺陷检测***、方法、装置、电子设备及存储介质”的中国专利申请的优先权,其全部内容通过引用结合在本申请中。
本申请实施例涉及机器视觉技术领域,特别是涉及一种缺陷检测***、方法、装置、电子设备及存储介质。
PVC(polyvinylchloride,聚氯乙烯)手套、丁腈手套等一次性手套在医疗、工业、食品等行业都有很大的需求,在满足市场需求的同时,PVC手套、丁腈手套等一次性手套的质量问题对于各行业的安全生产工作来说也是很重要的。因此,以PVC手套为例,生产PVC手套的生产商在着力研究如何高效率地生产PVC手套的同时,更注重生产出来的PVC手套的质量检测。为了实现高质量、高效率的生产活动,生产商纷纷选择使用机器视觉检测解决方案,以最大限度地缩短停机时间,确保始终提供安全、优质的产品。
相关技术中,采用单面视觉检测的方式,对PVC手套的缺陷进行检测,具体而言:流水线上自然下垂的PVC手套会在铰链带动下从用于间隔多排PVC手套的背景板前侧穿过,在通过背景板前时进入视觉检测区域;图像采集机构对进入视觉检测区域的一排PVC手套的一面(正面或背面)进行图像采集,并将采集到的手套图像上报控制装置,从而,控制装置根据手套图像对手套的缺陷情况进行检测。
然而,由于PVC手套的正反两面均有可能存在缺陷,而相关技术针对流水线上的一排手套,仅仅能够拍摄到一面,因此,相关技术中图像采集机构存在视野盲区,手套的缺陷检测结果不完整,最终导致得到的缺陷检测结果的准确率不高。
可见,针对流水线上的存在多面检测需求的产品对象,例如PVC手套、丁腈手套等一次性手套,如何提升缺陷检测的准确率,是一个亟待解决的问题。
发明内容
本申请实施例的目的在于提供一种缺陷检测***、方法、装置、电子设备及存储介质,以实现提高产品对象缺陷的检测结果的准确率。具体技术方案如下:
第一方面,为了达到上述目的,本申请实施例公开了一种缺陷检测***,包括:控制装置、光检装置和接近传感器;其中,上述光检装置设置有错位排布的第一相机和第二相机,上述第一相机为对流水线传递的待检测的每一产品对象的一面进行拍照的相机,上述第二相机为对每一产品对象的另一面进行拍照的相机;
上述接近传感器,设置在上述流水线上,设置为每当产品对象接近时,向上述控制装置发送感应信号;
上述控制装置,设置为每当接收到上述感应信号时,针对当前接近的产品对象进行编码,得到当前接近的产品对象的编码信息,以及向上述光检装置发送拍照触发信号;
上述光检装置,设置为响应于接收到的拍照触发信号,控制上述第一相机和第二相机进行拍照,得到第一图像和第二图像,并上报上述第一图像和第二图像至上述控制装置;
上述控制装置,还设置为接收各个第一图像和各个第二图像;以及,针对每一编码信息,按照预定的图像选取方式,选取包含该编码信息所指示的产品对象的第一图像和第二图像,并对所选取的第一图像以及第二图像进行产品对象缺陷检测处理,基于上述产品对象缺陷检测处理所得到的结果,确定该编码信息所指示的产品对象的缺陷检测结果;
其中,上述图像选取方式为基于上述接近传感器与第一相机之间的第一位置差异以及上述接近传感器与第二相机之间的第二位置差异所设定的选取方式。
第二方面,为了达到上述目的,本申请实施例公开了一种基于缺陷检测***的缺陷检测方法,应用于控制装置;上述方法包括:
每当接收到接近传感器发送的感应信号时,针对当前接近的产品对象进行编码,得到当前接近的产品对象的编码信息,以及向上述光检装置发送拍照触发信号,以使光检装置响应于接收到的拍照触发信号,控制上述第一相机和第二相机进行拍照,得到第一图像和第二图像,并上报上述第一图像和第二图像至上述控制装置;其中,上述接近传感器每当产品对象接近时向控制装置发送感应信号;
接收各个第一图像和各个第二图像;
针对每一编码信息,按照预定的图像选取方式,选取包含该编码信息所指示的产品对象的第一图像和第二图像,并对所选取的第一图像以及第二图像进行产品对象缺陷检测处理,基于上述产品对象缺陷检测处理所得到的结果,确定该编码信息所指示的产品对象的缺陷检测结果;
其中,上述图像选取方式为基于上述接近传感器与第一相机之间的第一位置差异以及上述接近传感器与第二相机之间的第二位置差异所设定的选取方式。
第三方面,为了达到上述目的,本申请实施例公开了一种基于缺陷检测***的缺陷检测装置,应用于控制装置;上述缺陷检测装置包括:
发送模块,设置为每当接收到接近传感器发送的感应信号时,针对当前接近的产品对象进行编码,得到当前接近的产品对象的编码信息,以及向上述光检装置发送拍照触发信号,以使光检装置响应于接收到的拍照触发信号,控制上述第一相机和第二相机进行拍照,得到第一图像和第二图像,并上报上述第一图像和第二图像至上述控制装置;其中,上述接近传感器每当产品对象接近时向控制装置发送感应信号;
接收模块,设置为接收各个第一图像和各个第二图像;
检测模块,设置为针对每一编码信息,按照预定的图像选取方式,选取包含该编码信息所指示的产品对象的第一图像和第二图像,并对所选取的第一图像以及第二图像进行产品对象缺陷检测处理,基于上述产品对象缺陷检测处理所得到的结果,确定该编码信息所指示的产品对象的缺陷检测结果;其中,上述图像选取方式为基于上述接近传感器与第一相机之间的第一位置差异以及上述接近传感器与第二相机之间的第二位置差异所设定的选取方式。
本申请实施例还提供了一种电子设备,包括处理器、通信接口、存储器和通信总线,其中,处理器,通信接口,存储器通过通信总线完成相互间的通信;存储器,用于存放计
算机程序;处理器,用于执行存储器上所存放的程序时,实现上述缺陷检测方法的步骤。
本申请实施例还提供了一种计算机可读存储介质,上述计算机可读存储介质内存储有计算机程序,上述计算机程序被处理器执行时实现上述缺陷检测方法的步骤。
本申请实施例还提供了一种包含指令的计算机程序产品,当其在计算机上运行时,使得计算机执行上述缺陷检测方法的步骤。
本身实施例还提供了一种包含指令的计算机程序,上述计算机程序在计算机上运行时,使得计算机执行上述缺陷检测方法的步骤。
本申请实施例有益效果:
本申请实施例提供了一种缺陷检测***,该***包括控制装置、光检装置和接近传感器,每当产品对象接近时,所述接近传感器向所述控制装置发送感应信号;所述控制装置每当接收到感应信号时,都会针对当前接近的产品对象进行编码,得到编码信息,然后向所述光检装置发送拍照触发信号;所述光检装置会响应于触发信号,控制第一相机和第二相机拍照,得到第一图像和第二图像;所述控制装置会获取各个第一图像和各个第二图像,针对每一编码信息,在预定的图像选取方式下,选取到对应产品对象的第一图像和第二图像,所述控制装置再对图像进行产品对象缺陷检测处理。
综上所述,本申请实施例中,光检装置获取到的流水线上产品对象的图像是包括了产品对象多面的图像,并且,控制装置可以基于预定的图像选取方式确定出关于每一产品对象对应的图像,即包含任一编码信息所指示的产品对象的第一图像和第二图像,进而对关于每一产品对象的图像进行产品对象缺陷检测处理,基于产品对象缺陷检测处理得到的结果,可以确定对应产品对象的关于双面的缺陷检测结果。可见,相对于现有技术采集单面图像进行缺陷检测处理而言,本方案的缺陷检测***基于多面视觉缺陷检测以及对各面对应的缺陷检测结果进行关联判定,来确定产品对象的缺陷检测结果,这样可以不存在视野盲区,缺陷检测结果的完整性大大提升,因此,可以提高产品对象的缺陷检测结果的准确率。
另外,本方案中,光检装置设置有错位排布的第一相机和第二相机,以对产品对象的多面进行图像采集。可见,通过该种光检装置可以在保证在线全检测的前提下,通过紧凑的设计空间,实现同时对产品对象的多面的图像采集,相比较传统的每一面单独检测的视觉检测***,大大提高了检测效率。
并且,本方案中,基于接近传感器,实现对于产品对象的编码,可以为同一产品对象的多面检测的关联提供实现条件。
当然,实施本申请的任一产品或方法并不一定需要同时达到以上所述的所有优点。
此处所说明的附图用来提供对本申请的进一步理解,构成本申请的一部分,本申请的示意性实施例及其说明用于解释本申请,并不构成对本申请的不当限定。
图1为本申请实施例提供的一种缺陷检测***的结构示意图;
图2为本申请实施例提供的另一种缺陷检测***的结构示意图;
图3(a)为本申请实施例提供的一种缺陷检测***中的光检装置示意图;
图3(b)、图3(c)、图3(d)和图3(e)为本申请实施例提供的一种缺陷检测***中的光检装置的各个视面的示意图;
图4为本申请实施例提供的另一种缺陷检测***的结构示意图;
图5为本申请实施例提供的另一种缺陷检测***的结构示意图;
图6为本申请实施例提供的一种缺陷检测***的设备整体主视图;
图7为本申请实施例提供的一种缺陷检测***的设备整体俯视图;
图8为本申请实施例提供的一种应用于控制装置的缺陷检测方法的流程示意图;
图9为本申请实施例提供的一种缺陷检测过程的流程示意图;
图10为本申请实施例提供的一种缺陷检测装置的装置示意图;
图11为本申请实施例提供的一种电子设备的结构示意图。
为使本申请的目的、技术方案、及优点更加清楚明白,以下参照附图并举实施例,对本申请进一步详细说明。显然,所描述的实施例仅仅是本申请一部分实施例,而不是全部的实施例。基于本申请中的实施例,本领域普通技术人员所获得的所有其他实施例,都属于本申请保护的范围。
为了方便理解方案,首先对本申请实施例所涉及的专业术语进行介绍:
机器视觉在线检测***:机器视觉在线检测***是集成在生产工艺流程内的检测设备,是集光学、机械、电子、计算、软件等技术为一体的工业应用***,它通过对电磁辐射的时空模式进行探测及感知,可以自动获取一幅或多幅目标物体图像,对所获取图像的各种特征量进行处理、分析,根据分析结果做出定性分析和定量解释得到相关目标物体的认识,并作出相应决策,是做出决策并执行实用的动作的过程控制***。
工业相机:机器视觉***中的一个关键组件,其最本质的功能就是将光信号转变成为有序的电信号。
接近传感器:是代替限位开关等接触式检测方式,以无需接触检测对象进行检测为目的的传感器。接近传感器能检测对象的移动信息和存在信息转换为电气信号。在转换为电气信号的检测方式中,包括利用电磁感应引起的检测对象的金属体中产生的涡电流的方式、捕测金属体的接近引起的电气信号的容量变化的方式、利石和引导开关的方式。
光检剔除机设备:一种利用光学成像、软件、电气等技术,集成手模和手套图像采集、缺陷检测及剔除机构的视觉***设备。
深度学习图像分割:基于深度学习卷积神经网络模型,把图像分成若干个特定的、具有独特性质的区域并提出感兴趣目标的技术和过程。它是由图像处理到图像分析的关键步骤。
深度学习目标检测:就是基于深度学习卷积神经网络模型,识别出图像中目标的类别,或者还要预测物体的位置,或针对多个目标确定它们的位置和类别。
随着市场需求的发展,缺陷检测***在工业化生产领域扮演着越来越重要的角色。相关技术中,在对产品对象进行缺陷检测时,存在人工检测方案,以及机器视觉在线检测方案。
以PVC手套、丁腈手套等一次性手套为例,现有的手套流水线中,手套在成型后,会在手模上随着流水线进入脱模工序。在此期间,需要检测手套表面是否有油污点、缺陷或破损等表面缺陷。
那么,以PVC手套的缺陷检测为例,相关技术所提供的人工检测方式如下:
在PVC手套在手模上随着流水线进入脱模工序的期间,采用人眼来检测PVC手套的表面缺陷,然后经人力将表面存在缺陷的手套从手模上剥除;这样,不仅需要较多的人力、劳动强度很大,而且检测准确率也较低,同时由于成本限制无法做到生产的手套全部检测,存在一定的遗漏概率。
以PVC手套的缺陷检测为例,相关技术所提供的机器视觉在线检测方案如下:
相关技术中,采用单面视觉检测的方式,对PVC手套的缺陷进行检测,具体而言:流水线上自然下垂的PVC手套会在铰链带动下从用于间隔多排PVC手套的背景板前侧穿过,在通过背景板前时进入视觉检测区域;图像采集机构对进入视觉检测区域的一排PVC手套的一面(正面或背面)进行图像采集,并将采集到的手套图像上报控制装置,从而,控制装置根据手套图像对手套的缺陷情况进行检测。
然而,由于PVC手套的正反两面均有可能存在缺陷,而相关技术针对流水线上的一排手套,仅仅能够被拍摄到一面,因此,相关技术中图像采集机构存在视野盲区,手套的缺陷检测结果不完整,最终导致得到的缺陷检测结果的准确率不高。
基于上述内容可知,相关技术中采用单面视觉检测的方式,对PVC手套、丁腈手套等一次性手套的缺陷进行检测,会带来缺陷检测结果的准确率不高的问题。因此,针对流水线上的存在多面检测需求的产品对象,例如PVC手套、丁腈手套等一次性手套,如何提升缺陷检测的准确率,是一个亟待解决的问题。
为了解决产品对象的缺陷检测结果的准确率不高的问题,本申请实施例提供了一种缺陷检测***、方法、装置、电子设备及存储介质。
其中,本申请提供了一种缺陷检测***,所述***包括控制装置、光检装置和接近传感器;其中,所述光检装置设置有错位排布的第一相机和第二相机,所述第一相机为对流水线传递的待检测的每一产品对象的一面进行拍照的相机,所述第二相机为对每一产品对象的另一面进行拍照的相机;上述***包括:
所述接近传感器,设置在所述流水线上,用于每当产品对象接近时,向所述控制装置发送感应信号;
所述控制装置,用于每当接收到所述感应信号时,针对当前接近的产品对象进行编码,得到当前接近的产品对象的编码信息,以及向所述光检装置发送拍照触发信号;
所述光检装置,用于响应于接收到的拍照触发信号,控制所述第一相机和第二相机进行拍照,得到第一图像和第二图像,并上报所述第一图像和第二图像至所述控制装置;
所述控制装置,还用于接收各个第一图像和各个第二图像;以及,针对每一编码信息,按照预定的图像选取方式,选取包含该编码信息所指示的产品对象的第一图像和第二图像,并对所选取的第一图像以及第二图像进行产品对象缺陷检测处理,得到该编码信息所指示的产品对象的缺陷检测结果;
其中,所述图像选取方式为基于所述接近传感器与第一相机之间的第一位置差异以及所述接近传感器与第二相机之间的第二位置差异所设定的选取方式。
需要说明的是,控制装置可以为与光检装置、接近传感器相通信的单独的控制设备,此时,所述控制设备能够实现所述控制装置所具有的控制功能。示例的,控制装置可以为部署有视觉检测软件的计算机;当然,所述控制装置所具有的功能也可以由至少两个设备相互配合来实现,也就是说控制装置的设备形态可以为至少两个设备的组合形态,这也是合理的。其中,视觉检测软件可以对接近传感器发出的感应信号进行编码、发出拍照触发信号、接收光检装置发送的图像、选取包含产品对象在内的图像,以及对图像做产品对象缺陷检测处理,并作出相应决策。
示例性的,光检装置中设置的相机可以是工业相机。本申请中光检装置可以借助工业相机的本质功能:将光信号转变成为有序的电信号,以实现接收到拍照触发信号后,进行拍照操作,得到图像,再将图像上报给控制装置。
接近传感器是代替限位开关等接触式检测方式,以无需接触检测对象进行检测为目的的传感器。
另外,产品对象可以是流水线上任一需要图像检测来获取缺陷检测结果的产品,示例性的,产品对象可以是PVC手套,或者丁腈手套等一次性手套,当然并不局限于此。
需要说明的是,本申请实施例中的控制装置、光检装置和接近传感器可以共同组成一套完整的机器视觉在线检测***,以实现在生产工艺流程中对产品进行缺陷检测处理,并基于缺陷检测结果作出相应决策。
综上所述,本申请实施例中,光检装置获取到的流水线上产品对象的图像是包括了产品对象多面的图像,并且,控制装置可以基于预定的图像选取方式确定出关于每一产品对象对应的图像,即包含任一编码信息所指示的产品对象的第一图像和第二图像,进而对关于每一产品对象的图像进行产品对象缺陷检测处理,基于产品对象缺陷检测处理得到的结果,可以确定对应产品对象的关于双面的缺陷检测结果。可见,相对于现有技术采集单面图像进行缺陷检测处理而言,本方案的缺陷检测***基于多面视觉缺陷检测以及对各面对应的缺陷检测结果进行关联判定,来确定产品对象的缺陷检测结果,这样可以不存在视野盲区,缺陷检测结果的完整性大大提升,因此,可以提高产品对象的缺陷检测结果的准确率。
另外,本方案中,光检装置设置有错位排布的第一相机和第二相机,以对产品对象的多面进行图像采集。可见,通过该种光检装置可以在保证在线全检测的前提下,通过紧凑的设计空间,实现同时对产品对象的多面的图像采集,相比较传统的每一面单独检测的视觉检测***,大大提高了检测效率。
并且,本方案中,基于接近传感器,实现对于产品对象的编码,可以为同一产品对象的多面检测的关联提供实现条件。
下面结合附图介绍本申请实施例所提供的一种缺陷检测***。
图1为本申请实施例提供的一种缺陷检测***的结构示意图,***包括:接近传感器110,控制装置120,光检装置130,接近传感器110与控制装置120相连,控制装置120
与光检装置130相连,所述光检装置130设置有错位排布的第一相机1301和第二相机1302,所述第一相机1301为对流水线传递的待检测的每一产品对象的一面进行拍照的相机,所述第二相机1302为对每一产品对象的另一面进行拍照的相机。其中,
所述接近传感器110,设置在所述流水线上,设置为每当产品对象接近时,向所述控制装置120发送感应信号。
所述控制装置120,设置为每当接收到所述感应信号时,针对当前接近的产品对象进行编码,基于所述产品对象缺陷检测处理所得到的结果,确定当前接近的产品对象的编码信息,以及向所述光检装置130发送拍照触发信号。
所述光检装置130,设置为响应于接收到的拍照触发信号,控制所述第一相机1301和第二相机1302进行拍照,得到第一图像和第二图像,并上报所述第一图像和第二图像至所述控制装置120。
所述控制装置120,还设置为接收各个第一图像和各个第二图像;以及,针对每一编码信息,按照预定的图像选取方式,选取包含该编码信息所指示的产品对象的第一图像和第二图像,并对所选取的第一图像以及第二图像进行产品对象缺陷检测处理,得到该编码信息所指示的产品对象的缺陷检测结果;
其中,所述图像选取方式为基于所述接近传感器110与第一相机1301之间的第一位置差异以及所述接近传感器110与第二相机1302之间的第二位置差异所设定的选取方式。
需要说明的是,本申请提供的缺陷检测***是集成在生产工艺流程内的检测设备,经过生产工艺后,流水线上已经有成排的产品对象等待进入缺陷检测***的检测区域,例如:以PVC手套为例,流水线上存在自然下垂的需要传递的两排或四排PVC手套,那么,每一排PVC手套会在铰链带动下进入缺陷检测***的检测区域。当然,针对流水线上存在的PVC手套的排数,本申请并不限定。并且,由于流水线上可以存在至少一排产品对象进行传递,因此,光检装置中针对每一排产品对象,可以设置有错位排布的第一相机和第二相机,也就是说,第一相机和第二相机针对同一排产品对象中每一产品对象的多面进行图像采集。
另外,沿着流水线的产品对象的传递方向,光检装置130的放置位置可以在接近传感器之后,也就是,任一产品对象先经过接近传感器,然后再传递一段时间后,经过光检装置130的图像采集区域,此时,光检装置130的相机拍摄到的产品对象,不同于接近传感器130感应到的产品对象。当然,光检装置130的放置位置也可以为满足如下条件的位置:任一产品对象靠近接近传感器110时,同时进入到光检装置的图像采集区域,此时,靠近接近传感器的产品对象,与光检装置中的目标相机拍摄到的产品对象,属于同一产品对象;其中,目标相机为第一相机和第二相机中靠近接近传感器130的相机。为了方便描述,下文中以第一相机作为靠近接近传感器130的相机进行方案介绍。
针对接近传感器110而言,由于流水线上设置有接近传感器,这样使得:每当产品对象接近时,接近传感器110都可以感应到,并向控制装置120发送感应信号,为控制装置120对产品对象进行编码的操作提供了条件。可以理解的是,接近传感器110的内部可以产生一个交变磁场,当悬挂在流水线的铰链上的产品对象随着流水线运行靠近时,铰链会与振动器产生的交变磁场发生电磁感应,以产生感应电流;在产品对象经过接近传感器110
的过程中,接近传感器110可以在不接触产品对象的前提下,将产品对象的位置信息和存在信息转换为感应信号,并将感应信号向控制装置120发送。其中,接近传感器110的种类选择可以根据生产现场需求而定,本申请并不对接近传感器110的具体种类进行限定。
针对控制装置120而言,每当接近传感器110发送感应信号时,控制装置120都可以接收到该感应信号,并对当前接近的产品对象进行编码,从而对先后经过接近传感器110的每一个位置上的产品对象分配标识信息,通过编码信息可以区分流水线上的不同位置上的产品对象,最终为同一产品对象的多面检测的关联提供实现条件。
并且,由于需要基于图像分析来实现缺陷检测的检测,因此,控制装置120每当接收到感应信号时,可以向所述光检装置130发送拍照触发信号,以触发光检装置中的第一相机和第二相机进行图像采集。也就是,每一次产品对象接近时,光检装置130均会被触发进行图像采集。
针对光检装置130而言,光检装置也可以称为光检机;光检装置130每当接收到控制装置120的拍照触发信号时,可以控制所述第一相机1301和第二相机1302进行拍照,得到第一图像和第二图像,并上传第一图像和第二图像至控制装置120。可以理解的是,由于第一相机1301和第二相机1302错位排布,第一相机1301和第二相机1302是在同一拍照触发信号触发下同时拍摄,所以得到的图像不可能是对应同一产品对象;但正是由于第一相机1301和第二相机1302错位排布,且第一相机和第二相机用于对同一排上的对象的不同面进行拍摄,因此,光检装置130在当前拍照触发信号作用下,可以触发第一相机1301拍出不被遮挡的产品对象的一面,也就是第一图像,而在另一拍照触发信号到来时,可以触发第二相机1302拍出不被遮挡的该产品对象的另一面,也就是第二图像,例如,若第一相机和第二相机对相邻产品对象进行图像采集,则该另一拍照触发信号,即为当前拍照触发信号的下一拍照触发信号。
可以理解的是,第一图像可以是被拍摄的产品对象的正面,相应的,第二图像可以是被拍摄的产品对象的背面;当然,第一图像也可以是被拍摄的产品对象的背面,相应的,第二图像可以是被拍摄的产品对象的正面。例如,在手套的生产流水线中,目标手套已经进入光检装置,光检装置响应于第一次触发信号,控制第一相机和第二相机拍照,第一相机可以拍摄到目标手套的正面,第二相机可以拍到不包含目标手套在内的图像;随着流水线运行,光检装置响应于第二次触发信号,控制第一相机和第二相机拍照,第一相机可以拍摄到不包含目标手套在内的图像,第二相机可以拍到目标手套的背面。
另外,可以理解的是,在生产现场,接近传感器110与第一相机1301之间的第一位置差异以及接近传感器110与第二相机1302之间的第二位置差异是一种规律特征,可以基于该规律特征,来为每一产品对象关联到包含该产品对象的第一图像和第二图像。具体而言,控制装置120通过编码信息可以标识出流水线上依次靠近接近传感器130的各个产品对象,从而结合已经确定的第一位置差异和第二位置差异,可以选取到对应各个产品对象的第一图像和第二图像,也就是,按照预定的图像选取方式,选取包含每一编码信息所指示的产品对象的第一图像和第二图像。例如,接近传感器、第一相机和第二相机紧密排布,各部件之间没有间隔,即不存在产品对象,第一产品对象、第二产品对象和第三产品对象是按时间先后顺序依次靠近接近传感器的产品对象,第一产品对象、第二产品对象和第三产品
对象是流水线上的相邻的产品对象;那么,第一产品对象靠近接近传感器时,第一相机和第二相机拍摄到不包含第一产品对象的图像;第二产品对象靠近接近传感器时,第一相机拍摄到第一产品对象的正面图像,第二相机拍摄到不包含任一产品对象的图像;第三产品对象靠近接近传感器时,第一相机拍摄到第二产品对象的正面图像,第二相机拍摄到第一产品对象的背面图像;至此,控制装置可以基于位置差异和编码信息,选取到第一产品对象的正面图像和背面图像。又例如,一产品对象可以靠近接近传感器的同时,进入第一相机的图像采集区域,第一相机和第二相机拍摄相邻的产品对象,这种情况下,一产品对象靠近接近传感器时,接近传感器发出信号而导致控制装置发出拍照触发信号时,第一相机可以拍摄到正在靠近接近传感器的该产品对象,第二相机拍不到,而该产品对象的下一产品对象接近时,第二相机才可以拍摄到之前经过接近传感器的产品对象的图像,即该产品对象的图像,此时第一相机拍摄到的第一张图像以及第二相机拍摄的第二张图像,可以作为该产品对象的待关联的两个图像。
控制装置120可以在选取到对应产品对象的第一图像和第二图像后,对图像进行产品对象缺陷检测处理,由此就可以得到对应产品对象的缺陷检测结果。示例性的,控制装置可以使用预先训练用于进行产品对象缺陷检测的检测模型,然后将第一图像和第二图像输入该检测模型以进行产品对象缺陷检测。其中,缺陷检测模型的建立可以借助深度学习分割、目标检测技术实现。例如,产品对象缺陷检测的检测模型可以采用深度学习与传统2D算法相结合的方案,根据人眼判定的缺陷种类加入网络中学习,以此训练得到模型。本申请实施例并不对具体的缺陷检测过程进行限定。
其中,基于所述产品对象缺陷检测处理所得到的结果,确定该编码信息所指示的产品对象的缺陷检测结果的确定方式可以存在多种。示例性的,控制装置120在对第一图像和第二图像进行产品对象缺陷检测处理后,可以分别获得第一图像和第二图像的缺陷检测结果;当第一图像和第二图像的缺陷检测结果都表征产品对象有缺陷时,也就是,在与的逻辑关系下,可以得到对应产品对象有缺陷的结果,即该编码信息所指示的产品对象的缺陷检测结果为有缺陷;或者,当第一图像或第二图像的缺陷检测结果中至少有一个表征产品对象有缺陷时,也就是,在或的逻辑关系下,就可以得到对应产品对象有缺陷的结果,即该编码信息所指示的产品对象的缺陷检测结果为有缺陷;或者,当第一图像和第二图像的缺陷检测结果都表征产品对象没有缺陷时,也就是,在非的逻辑关系下,可以得到对应产品对象没有缺陷的结果,即该编码信息所指示的产品对象的缺陷检测结果为没有缺陷。
另外,第一图像的缺陷检测结果的具体内容和第二图像的缺陷结果的具体内容可以是用于表示是否有缺陷的文本内容,例如:“有缺陷”或“无缺陷”,此时,若一图像的缺陷检测结果为“有缺陷”则表征是产品对象的一面有缺陷,而若一图像的缺陷检测结果为“无缺陷”,则表征产品对象的一面没有缺陷。当然,第一图像的缺陷检测结果的具体内容和第二图像的缺陷结果的具体内容可以为缺陷类别信息,其中,缺陷类别信息可以包括用于表征没有缺陷的类别信息,以及在存在缺陷时该缺陷所属的具体缺陷类别的类别信息;此时,若一图像的缺陷检测结果为具体缺陷类别的类别信息,则表征是产品对象的一面有缺陷,而若一图像的缺陷检测结果为表征没有缺陷的类别信息,则表征产品对象的一面没有缺陷。需要说明的是,第一图像的缺陷检测结果的具体内容和第二图像的缺陷结果的具体
内容可以为缺陷类别信息的情况下,该编码信息所指示的产品对象的缺陷检测结果中还可以表征出所具有的缺陷的具体类别信息。
可选的,光检装置130可以通过频闪的光源控制,增加了成像的亮度范围,大大提升成像的质量,使成像更接近人眼感知,并且可以兼容高速生产线节拍,同时增加硬件的使用寿命。
另外,流水线的用于传递产品对象的支架上还可以设置有限位器。
针对限位器而言,每当产品对象接近接近传感器时,限位器可以使流水线平稳短暂地停止,此时光检装置可以控制相机拍照,得到清晰的图像。
综上所述,本申请实施例中,光检装置获取到的流水线上产品对象的图像是包括了产品对象多面的图像,并且,控制装置可以基于预定的图像选取方式确定出关于每一产品对象对应的图像,即包含任一编码信息所指示的产品对象的第一图像和第二图像,进而对关于每一产品对象的图像进行产品对象缺陷检测处理,基于产品对象缺陷检测处理得到的结果,可以确定对应产品对象的关于双面的缺陷检测结果。可见,相对于现有技术采集单面图像进行缺陷检测处理而言,本方案的缺陷检测***基于多面视觉缺陷检测以及对各面对应的缺陷检测结果进行关联判定,来确定产品对象的缺陷检测结果,这样可以不存在视野盲区,缺陷检测结果的完整性大大提升,因此,可以提高产品对象的缺陷检测结果的准确率。
另外,本方案中,光检装置设置有错位排布的第一相机和第二相机,以对产品对象的多面进行图像采集。可见,通过该种光检装置可以在保证在线全检测的前提下,通过紧凑的设计空间,实现同时对产品对象的多面的图像采集,相比较传统的每一面单独检测的视觉检测***,大大提高了检测效率。
并且,本方案中,基于接近传感器,实现对于产品对象的编码,可以为同一产品对象的多面检测的关联提供实现条件。
可选的,在另一实施例中,如图2所示,在图1所示缺陷检测***的基础上,每一产品对象在流水线上传递时,关联有一产品模具;所述光检装置130还设置有错位排布的第三相机1303和第四相机1304,所述第三相机1303为对流水线传递的各个产品模具的一面进行拍照的相机,所述第四相机1304为对各个产品模具的另一面进行拍照的相机。
所述光检装置130,还设置为响应于接收到的拍照触发信号,控制所述第三相机1303和第四相机1304进行拍照,得到第三图像和第四图像,并向所述控制装置120上报所述第三图像和第四图像。
所述控制装置120,还设置为接收各个第三图像和各个第四图像;以及,针对每一编码信息,确定与所选取的第一图像的拍摄时间相匹配的第三图像,以及与所选取的第二图像的拍摄时间相匹配的第四图像,对所选取的第三图像和第四图像进行产品模具缺陷检测处理,基于所述产品模具缺陷检测处理所得到的结果,确定模具检测结果,将所得到的模具检测结果与该编码信息所指示的产品对象的缺陷检测结果进行关联。
由于产品模具通常在产品对象的上方,例如:手模下方关联着手套,因此,第三相机可以位于第一相机的上方,第四相机位于第二相机的上方,这样使得第一相机对经过的产
品对象进行拍照时,该经过的产品对象的产品模具可以同时进入第三相机的图像采集区域,类似的,第二相机对经过的产品对象进行拍照时,该经过的产品对象关联的产品模具可以同时进入第四相机的图像采集区域。另外,由于流水线上可以存在至少一排产品对象进行传递,那么,产品模具可以存在至少一排,因此,光检装置中针对每一排产品对象,可以设置有错位排布的第一相机和第二相机,以及针对每一排产品模具,设置有错位排布的第三相机和第四相机,也就是说,第一相机和第二相机针对同一排产品对象中每一产品对象的多面进行图像采集,第三相机和第四相机针对同一排产品模具中每一产品模具的多面进行图像采集。示例性的,在具体应用中,包含第一相机1301、第二相机1302、第三相机1303和第四相机1304的光检装置的示意图可以如图3(a)所示;其中,虚线框内区域为相机镜头光路拍摄范围,第一相机1301、第二相机1302、第三相机1303和第四相机1304安装在光检装置设备主体1305;为了更清楚地示出光检装置,图3(b)-图3(e)示出了光检装置的各个视面的工程示意图。
需要说明的是,光检装置130中的第三相机1303和第四相机1304所实现的功能与第一相机1301和第二相机1302可以基本一致,不同的是,第三相机1303和第四相机1304拍摄对象可以为与产品对象关联的产品模具;相应的,控制装置120可以确定与所选取的第一图像的拍摄时间相匹配的图像为第三图像,以及确定与所选取的第二图像的拍摄时间相匹配的图像为第四图像。例如,在手套生产流水线,选定一特定手套,已经确定了对应特定手套的第一图像,那么可以选取与拍摄该第一图像的第一相机同时拍摄的第三相机拍摄到的第三图像,该第三图像就是与特定手套关联的特定手模的图像;同理可以选取到包含与特定手套关联的特定手模的第四图像。
可以理解的是,在控制装置120对所确定的第三图像和第四图像进行缺陷检测处理时,缺陷检测处理的方式与控制装置120对所确定的第一图像和第二图像进行缺陷检测处理的方式一致,在本实施例中就不做赘述。
需要说明的是,基于所述产品模具缺陷检测处理所得到的结果,确定模具检测结果的确定方式可以存在多种。相应的,基于所述产品对象缺陷检测处理所得到的结果,确定该编码信息所指示的产品对象的缺陷检测结果的确定方式也可以存在多种。可以理解的是,在与、或、非三种逻辑关系的情况下,控制装置120可以依据第一图像和第二图像的缺陷检测结果,确定产品对象的缺陷检测结果;其中,具体确定方式已在上一实施例中描述;相应的,在与、或、非三种逻辑关系的情况下,控制装置120也可以依据第三图像和第四图像的缺陷检测结果,确定产品模具的缺陷检测结果。示例性的,控制装置120在对第三图像和第四图像进行产品模具缺陷检测处理后,可以分别获得第三图像和第四图像的缺陷检测结果;当第三图像和第四图像的缺陷检测结果都表征产品模具有缺陷时,也就是,在与的逻辑关系下,可以得到对应产品模具有缺陷的结果,即该编码信息所指示的产品模具的缺陷检测结果为有缺陷;或者,当第三图像或第四图像的缺陷检测结果中至少有一个表征产品模具有缺陷时,也就是,在或的逻辑关系下,就可以得到对应产品模具有缺陷的结果,即该编码信息所指示的产品模具的缺陷检测结果为有缺陷;或者,当第三图像和第四图像的缺陷检测结果都表征产品模具没有缺陷时,也就是,在非的逻辑关系下,可以得到对应产品模具没有缺陷的结果,即该编码信息所指示的产品模具的缺陷检测结果为没有缺
陷。
另外,第三图像的缺陷检测结果的具体内容和第四图像的缺陷结果的具体内容可以是用于表示是否有缺陷的文本内容,例如:“有缺陷”或“无缺陷”,此时,若一图像的缺陷检测结果为“有缺陷”则表征是产品模具的一面有缺陷,而若一图像的缺陷检测结果为“无缺陷”,则表征产品模具的一面没有缺陷。当然,第三图像的缺陷检测结果的具体内容和第四图像的缺陷结果的具体内容可以为缺陷类别信息,其中,缺陷类别信息可以包括用于表征没有缺陷的类别信息,以及在存在缺陷时该缺陷所属的具体缺陷类别的类别信息;此时,若一图像的缺陷检测结果为具体缺陷类别的类别信息,则表征是产品模具的一面有缺陷,而若一图像的缺陷检测结果为表征没有缺陷的类别信息,则表征产品模具的一面没有缺陷。
可以理解的是,控制装置120在得到产品模具的缺陷检测结果后,可以将模具缺陷检测结果与产品对象的缺陷检测结果关联。例如,在手套生产流水线,目标手模缺陷检测结果表明手模有裂纹的缺陷,垂在目标手模下方的目标手套的缺陷检测结果表明手套有破损、裂纹的缺陷;那么,***可以将手套破损、裂纹缺陷与手模有裂纹缺陷关联。又例如,在手套生产流水线,目标手模缺陷检测结果表明手模有脏污的缺陷,垂在目标手模下方的目标手套的缺陷检测结果表明手套有油污、流痕的缺陷;那么,***可以将手套油污、流痕缺陷与手模有脏污缺陷关联。
另外,在一种实现方式中,缺陷检测***还可以针对产品对象的缺陷进行分级处理,当控制装置检测到严重影响产品对象质量的缺陷时,控制装置可以将与产品对象缺陷关联的产品模具缺陷做公告通知处理,以提醒相关人员对产品模具做维保更换处理。例如,在手套生产流水线,控制装置检测出手套有破损缺陷,与该手套关联的手模缺陷检测结果表明手模有裂纹的缺陷;那么,***可以将手模有裂纹缺陷的信息发送给相关人员,以提醒相关人员对手模做维修处理。
另外,在另一种实现方式中,控制装置可以在检测出产品对象缺陷和产品模具缺陷后,保存产品对象缺陷和产品模具缺陷数据,并进行统计分析,这样可以在数据支持下,对产品对象缺陷和产品模具缺陷作出预测。
本申请实施例中,缺陷检测***可以同时对产品对象和产品模具进行缺陷检测,并将二者缺陷检测结果关联起来,这样可以进一步为缺陷成因的分析提供基础数据。
可选的,在另一实施例中,如图4所示,在图1所示缺陷检测***的基础上,控制装置120包括:主控装置410和可编程逻辑控制器(Programmable Logic Controller)PLC 420;
所述PLC 420,设置为每当接收到所述感应信号时,针对当前接近的产品对象进行编码,并将编码信息发送至主控装置410,以及向所述光检装置130发送拍照触发信号。
所述主控装置410,设置为接收各个第一图像和各个第二图像;以及,针对每一编码信息,按照预定的图像选取方式,选取包含该编码信息所指示的产品对象的第一图像和第二图像,并对所选取的第一图像以及第二图像进行产品对象缺陷检测处理,基于所述产品对象缺陷检测处理所得到的结果,确定该编码信息所指示的产品对象的缺陷检测结果。
所述光检装置130具体设置为上报所述第一图像和第二图像至所述主控装置410。
可以理解的是,在本实施例中,在控制装置120可以细分为主控装置410和可编辑逻辑控制器PLC420后,功能也相应划分。PLC 420可以接收每一感应信号,每一感应信号代表了每一接近的产品对象。因此,PLC可以对当前接近的产品对象进行编码,然后可以再将编码信息发送至主控装置410,以及向光检装置130发送拍照触发信号。而主控装置410可以接收各个第一图像和各个第二图像;以及针对每一编码信息,按照预定的图像选取方式,选取包含该编码信息所指示的产品对象的第一图像和第二图像,并对所选取的第一图像以及第二图像进行产品对象缺陷检测处理,得到该编码信息所指示的产品对象的缺陷检测结果。
其中,主控装置410和PLC420属于分离状态的设备,主控装置410可以是任一形态的电子设备,本申请实施例并不做限定。
通过该实施例所提供的方案,控制装置由主控装置410和PLC420构成,这样,通过主控装置410和PLC420的配合可以实现控制装置所具有的功能。通过该种分离形式,可以实现功能解耦,从而达到提升***维护性的便捷性、降低运营成本的目的。
可选的,在本申请的另一实施例中,控制装置120针对每一编码信息,按照预定的图像选取方式,选取包含该编码信息所指示的产品对象的第一图像和第二图像,可以包括步骤A1:
步骤A1,针对每一编码信息,基于该编码信息对应的次序、各个第一图像对应的次序以及所述第一位置差异,确定包含该编码信息所指示的产品对象的第一图像;以及,基于该编码信息对应的次序、各个第二图像对应的次序以及所述第二位置差异,确定包含该编码信息所指示的产品对象的第二图像;
其中,该编码信息对应的次序为按照生成时间顺序对各个编码信息进行排序时该编码信息对应的次序;每一第一图像对应的次序为按照拍摄时间对各个第一图像进行排序时,该第一图像对应的次序;每一第二图像对应的次序为按照拍摄时间对各个第二图像进行排序时,该第二图像对应的次序。
本实施例中,可以按照生成时间顺序,对各个编码信息进行排序,从而得到每一编码信息对应的次序,即每一编码信息在各个编码信息构成的队列中的排列位置;并且,可以按照拍摄时间对各个第一图像进行排序,从而得到每一第一图像对应的次序,即每一第一图像在各个第一图像构成的队列中的排列位置;类似的,可以按照拍摄时间对各个第二图像进行排序,从而得到每一第二图像对应的次序,即每一第二图像在各个第二图像构成的队列中的排列位置。
基于上述排列位置的确定,考虑到第一位置差异和第二位置差异的存在,会使得同一产品对象的编码信息、第一图像和第二图像,在各自所处的队列中的位置不同,且存在规律性,因此,可以针对每一编码信息,基于该编码信息对应的次序、各个第一图像对应的次序以及所述第一位置差异,确定包含该编码信息所指示的产品对象的第一图像;以及,基于该编码信息对应的次序、各个第二图像对应的次序以及所述第二位置差异,确定包含该编码信息所指示的产品对象的第二图像。通过该种方式,可以快速得到同一产品对象的第一图像和第二图像。
需要说明的是,按照生成时间顺序对各个编码信息进行排序得到的次序可以是一个有先后顺序的队列,队列内各个编码信息之间的时间间隔可以取决于生产现场流水线的运行速度;相应的,各个第一图像之间和各个第二图像之间的时间间隔与编码信息之间的时间间隔可以是一致的。
并且,编码信息排序得到的次序可以形成一个编码信息队列,各个第一图像排序得到的次序可以形成一个第一图像队列,各个第二图像排序得到的次序可以形成一个第二图像队列;这三个队列中元素之间的时间间隔可以是一致的,各元素在队列中的位置可以是已知的。在一个产品对象靠近接近传感器时,第一相机和第二相机可以拍照;相应的,编码信息队列中可以多一个关于该产品对象的编码信息元素,第一图像队列可以多一个第一相机拍摄到的图像元素,第二图像队列可以多一个第二相机拍摄到的图像元素。在已知接近传感器、第一相机和第二相机的位置差异的情况下,即,在已知接近传感器、第一相机和第二相机之间相差元素的数量时,是可以确定包含对应编码信息所指示的产品对象的第一图像和第二图像的。
可选的,在一种实现方式中,所述第一位置差异可以为:当第一相机的拍摄视野内进入待拍摄的产品对象且存在产品对象接近接近传感器时,进入第一相机的拍摄视野的产品对象与接近所述接近传感器的产品对象之间,间隔N1个产品对象;
所述第二位置差异可以为:当第二相机的拍摄视野内进入待拍摄的产品对象且存在产品对象接近接近传感器时,进入第二相机的拍摄视野的产品对象与接近所述接近传感器的产品对象之间,间隔N2个产品对象;其中,N1和N2为不同的数值。
相应的,所述控制装置基于该编码信息对应的次序、各个第一图像对应的次序以及所述第一位置差异,确定包含该编码信息所指示的产品对象的第一图像,包括:
从各个第一图像中,选取所对应次序不小于该编码信息对应的次序且相差N1+1的第一图像;
所述控制装置基于该编码信息对应的次序、各个第二图像对应的次序以及所述第二位置差异,确定包含该编码信息所指示的产品对象的第二图像,包括:
从各个第二图像中,选取所对应次序不小于该编码信息对应的次序且相差N2+1的第二图像。该种实现方式中,描述位置差异时,可以在相机视野中有待拍摄的产品对象且存在新的产品对象接近接近传感器时,将待拍摄产品对象和新的产品对象之间间隔的产品对象的个数当做距离单位。例如,在手套生产流水线中,第一相机与接近传感器之间间隔5个手套,0号手套进入第一相机拍摄视野,与此同时,新的手套接近接近传感器,那么新的手套可以记为6号手套。其中,N1和N2可以是非零自然数;特别的,当接近传感器与第一相机紧密连接时N1可以为零。该种实现方式中,在选取第一图像时,可以是在一组由各个第一图像组成的队列中选取第一图像,队列中各个第一图像按拍照的先后顺序排列;可以理解的是,排列顺序在先的第一图像在队列中的次序小,因此,选取所对应次序不小于该编码信息对应的次序且相差N1+1的第一图像,也就是说,选取的第一图像对应的次序,与编码信息对应的次序,相差N1+1,且第一图像对应的次序不小于编码信息对应的次序,即,选取第一相机在流水线从接近传感器开始运行了N1个产品对象后拍摄到的图像,也就是第N1+1个图像。例如,接近传感器与第一相机间隔5个产品对象,针对任一
产品对象,若该产品对象所具有的编码信息对应的次序为1,***会选取第一图像队列中的次序为6的图像,得到该产品对象的第一图像。
第二图像的选取规则与第一图像同理,区别在于,接近传感器和第一相机之间的间隔N1与接近传感器和第二相机之间的间隔N2不同。具体而言,可以选取所对应次序不小于该编码信息对应的次序且相差N2+1的第二图像,也就是说,选取的第二图像对应的次序,与编码信息对应的次序,相差N2+1,且第二图像对应的次序不小于编码信息对应的次序,即,选取第二相机在流水线从接近传感器开始运行了N2个产品对象后拍摄到的图像,也就是第N2+1个图像。
基于上述的第一位置差异和第二位置差异的描述方式,在控制装置120对于第一图像和第二图像的编码方式不同时,下述的步骤B1和步骤B2的具体实现方式不同。
示例性的,在一种实现方式中,控制装置120还可以设置为:从第一相机拍摄得到的首张第一图像开始,对各个第一图像按照拍摄时间进行次序编码,得到各个第一图像的编码值;以及从第二相机拍摄得到的首张第二图像开始,对各个第二图像按照拍摄时间进行次序编码,得到各个第二图像的编码值。也就是,从首张第一图像开始,对各个第一图像,按照次序进行编码,通过编码的方式表征各个第一图像的次序;从首张第二图像开始,对各个第二图像,按照次序进行编码,通过编码的方式表征各个第二图像的次序。
步骤B1,控制装置120从各个第一图像中,选取所对应次序不小于该编码信息对应的次序且相差N1+1的第一图像,可以包括步骤B11:
B11,从各张第一图像中,确定所具有编码值符合第一条件的第一图像,得到所对应次序不小于该编码信息对应的次序且相差N1+1的第一图像;其中,所述第一条件为所具有编码值不小于该编码信息对应的次序且相差N1+1。
步骤B2,控制装置120从各个第二图像中,选取所对应次序不小于该编码信息对应的次序且相差N2+1的第二图像,可以包括步骤B21:
B21,从各张第二图像中,确定所具有编码值符合第二条件的第二图像,得到所对应次序不小于该编码信息对应的次序且相差N2+1的第二图像;其中,所述第二条件为所具有编码值不小于该编码信息对应的次序且相差N2+1。
其中,第一图像和第二图像的编码值即为表征次序的值。
可以理解的是,从第一相机拍摄得到的首张第一图像开始,可以对各个第一图像按照拍摄时间进行次序编码,在这种情况下,第一相机拍摄到的各个第一图像中可以有不包含产品对象的图像。需要说明的是,选取对应产品对象的第一图像和第二图像的方法可以与步骤B1和步骤B2一致。
例如,在手套流水线,接近传感器、第一相机和第二相机之间间隔相同,流水线运行速度保持不变,每一手套靠近接近传感器的频率保持不变,第一相机和第二相机的拍照频率保持不变;每次拍照发生时,各个手套的位置如表1所示;其中,1号手套、2号手套和3号手套是按时间先后顺序靠近接近传感器的手套;第一相机所拍摄到的各个第一图像按照时间顺序可以被编码为1号图像、2号图像和3号图像;类似的,第二相机所拍摄到的各个第二图像按照时间顺序也可以被编码为1号图像、2号图像和3号图像;那么,1号手套靠近接近传感器时,第一相机拍摄到不包含手套图像的1号图像,第二相机拍摄到
不包含手套图像的1号图像;2号手套靠近接近传感器时,第一相机拍摄到包含1号手套正面图像的2号图像,第二相机拍摄到不包含手套图像的2号图像;3号手套靠近接近传感器时,第一相机拍摄到包含2号手套正面图像的3号图像,第二相机拍摄到包含1号手套背面图像的3号图像;此时,控制装置可以选取到第一相机拍摄到的2号图像和第二相机拍摄到的3号图像作为1号手套的正面图像和背面图像。
表1
示例性的,在另一种实现方式中,控制装置120还可以设置为:从第一相机拍摄得到的第N1+1张第一图像的下一图像开始,对各个第一图像按照拍摄时间进行次序编码,得到各个第一图像的编码值;以及从第二相机拍摄得到的第N2+1张第二图像的下一图像开始,对各个第二图像按照拍摄时间进行次序编码,得到各个第二图像的编码值。
步骤B1,控制装置120从各个第一图像中,选取所对应次序不小于该编码信息对应的次序且相差N1+1的第一图像,可以包括步骤B12:
B12,从各个第一图像中,确定所具有编码值与该编码信息对应的次序相匹配的第一图像,得到所对应次序不小于该编码信息对应的次序且相差N1+1的第一图像。
步骤B2,控制装置120从各个第二图像中,选取所对应次序不小于该编码信息对应的次序且相差N2+1的第二图像,可以包括步骤B22:
B22,从各个第二图像中,确定所具有编码值与该编码信息对应的次序相匹配的第二图像,得到所对应次序不小于该编码信息对应的次序且相差N2+1的第二图像。
可以理解的是,从第一相机拍摄得到的第N1+1张第一图像的下一图像开始,对各个第一图像按照拍摄时间进行次序编码,以及从第二相机拍摄得到的第N2+1张第二图像的下一图像开始,对各个第二图像按照拍摄时间进行次序编码,在这种情况下,第一相机和第二相机拍摄到的各个第一图像和各个第二图像可以是包含产品对象的图像。需要说明的是,对应产品对象的第一图像和第二图像的选取方法可以与步骤B1和步骤B2一致。
例如,在手套流水线,接近传感器与第一相机之间间隔1个相邻手套之间的距离;接近传感器与第二相机之间间隔2个相邻手套之间的距离,流水线运行速度保持不变,每一手套靠近接近传感器的频率保持不变,第一相机和第二相机的拍照频率保持不变;每次拍照发生时,各个手套的位置如表2所示;其中,1号手套、2号手套、3号手套、4号手套和5号手套是按时间先后顺序靠近接近传感器的手套;从第一相机拍摄得到的第3张图像开始,对第一相机所拍摄到的各个第一图像按照拍摄时间顺序编码为3号图像、4号图像和5号图像;类似的,从第二相机拍摄得到的第4张图像开始,对第二相机所拍摄到的各个第二图像按照拍摄时间顺序编码为4号图像、5号图像和6号图像;那么,3号手套靠近接近传感器时,第一相机拍摄到包含2号手套正面图像的3号图像,第二相机拍摄到包含1号手套背面图像的4号图像;4号手套靠近接近传感器时,第一相机拍摄到包含3号手套正面图像的4号图像,第二相机拍摄到包含2号手套背面图像的5号图像;5号手套
靠近接近传感器时,第一相机拍摄到包含4号手套正面图像的5号图像,第二相机拍摄到包含3号手套背面图像的6号图像;此时,控制装置可以选取到第一相机拍摄到的4号图像和第二相机拍摄到的6号图像作为3号手套的图像。
表2
本申请实施例中,控制装置可以针对拍摄到的包含产品对象的图像或不包含产品对象的图像,进行相应的编码,得到图像次序,这样可以基于产品对象编码信息对应的次序、各个图像对应的次序以及***中各部件位置差异,选取到对应产品对象的图像,以便于产品对象缺陷检测,从而进一步提高产品对象的缺陷检测结果准确率。
可选的,在另一实施例中,如图5所示,在图1所示缺陷检测***的基础上,还包括:剔除装置510,其中,所述剔除装置在所述流水线上的位置处于所述光检装置130之后;
所述控制装置120,还设置为在检测到任一编码信息所指示的产品对象的缺陷检测结果,符合剔除条件时,控制所述剔除装置510从流水线上剔除该编码信息所指示的产品对象。
所述剔除装置510,设置为在所述控制装置120的控制下,从流水线上剔除该编码信息所表示的对象。
其中,剔除装置也可以称为剔除机,通过剔除装置可以实现对于不合规的产品对象的剔除。
可选的,在一种实现方式中,接近传感器110与所述剔除装置510的位置差异为:当存在产品对象进入所述剔除装置的可操作区域且存在产品对象接近所述接近传感器时,进入所述可操作区域的产品对象与接近所述接近传感器的产品对象之间,间隔N3个产品对象。
控制装置控制所述剔除装置从流水线上剔除该编码信息所指示的产品对象,可以包括步骤:
在接收到指定感应信号之后,第N3+1次接收到感应信号时,向所述剔除装置发送对象剔除指令,以使所述剔除装置从流水线上剔除该编码信息所指示的产品对象;
其中,所述指定感应信号为该编码信息所表示的产品对象接近时所发出的感应信号。
可以理解的是,控制装置120在检测到产品对象的缺陷检测结果,符合剔除条件时,可以向剔除装置510发送剔除该产品对象的指令;剔除装置510可以接收到控制装置发送的指令,对第N3+1个产品对象执行剔除操作。例如,控制装置与剔除装置之间间隔6个产品对象,控制装置检测出0号产品对象需要进行剔除操作,在7号产品对象刚好靠近接近式传感器时,0号产品进入剔除装置的操作区域,因此,控制装置可以在接收到7号产品对象的感应信号时,发出剔除指令。
其中,剔除条件可以是确定产品对象的缺陷检测结果中,包括预先确定好的剔除缺陷。
例如,在手套生产流水线上,当控制装置检测出待测手套有断指、破洞、大撕破、腕部刮破、黑油污、黄油污、二次料、裂纹、流痕、余料的缺陷时,控制剔除装置从流水线上剔除该待测手套。本申请实施例并不对具体的剔除条件进行限定。
需要说明的是,剔除装置510可以应用在光检装置130包括第一相机、第二相机、第三相机和第四相机的场景。示例性的,在图2所示的实施例基础上,剔除装置510可以连接在光检装置之后,用于从流水线上剔除指定的产品对象。
示例性的,在手套生产流水线上,剔除装置可以包括设备主体和抓钩,抓钩可以由安置在设备主体上的可控制机械臂控制,在可控制机械臂接收到控制装置发送的剔除指令后,可以通过抓钩拉扯手套的方式对目标手套执行剔除操作。
本申请实施例中,在缺陷检测***中增加了剔除装置,并给控制装置配备了对应剔除装置的功能,这样在产品对象的生产流水线上可以及时清除不合格的产品对象,提高产品对象的质量。
为了方便理解缺陷检测***,下面结合图6对***进行示意图性说明。示例性的,如图6所示,本实施例提供的缺陷检测***应用于手套生产流水线,可以包括接近传感器(图6中未显示)、限位器01、光检装置130、剔除装置510和部署有控制装置的机柜04。相应的,图7为图6的俯视图。
示例性的,光检装置130可以获取到流水线上手套多面的图像,并且,控制装置可以基于预定的图像选取方式确定出关于每一手套对应的图像,即包含任一编码信息所指示的手套的第一图像和第二图像,进而对关于每一手套的图像进行手套缺陷检测处理,以得到手套的多面缺陷检测结果;控制装置可以依据手套缺陷检测结果,判断出需要剔除的手套,将剔除信号发送给剔除装置510;剔除装置510可以执行剔除动作,剔除掉需要剔除的手套。
示例性的,光检装置130和剔除装置510可以组合成光检剔除机设备。从而实现对手套图像获取以及缺陷手套剔除功能的耦合,提升缺陷检测以及缺陷剔除的效率。
本申请实施例提供的缺陷检测***属于一种机器视觉在线检测***,由接近传感器、限位器、光检装置、剔除装置和控制装置组成,该***不但可以实现检测分析结果及时反馈设备执行,也可以更好适配不同产线,实现灵活的拆分和链接。
基于上述的缺陷检测***,从控制装置的角度,本申请实施例提供了一种缺陷检测方法。如图8所示,该缺陷检测方法应用于缺陷检测***的控制装置,可以包括如下步骤:
S801,每当接收到接近传感器发送的感应信号时,针对当前接近的产品对象进行编码,得到当前接近的产品对象的编码信息,以及向所述光检装置发送拍照触发信号,以使光检装置响应于接收到的拍照触发信号,控制所述第一相机和第二相机进行拍照,得到第一图像和第二图像,并上报所述第一图像和第二图像至所述控制装置;其中,所述接近传感器每当产品对象接近时向控制装置发送感应信号。
S802,接收各个第一图像和各个第二图像。
S803,针对每一编码信息,按照预定的图像选取方式,选取包含该编码信息所指示的
产品对象的第一图像和第二图像,并对所选取的第一图像以及第二图像进行产品对象缺陷检测处理,得到该编码信息所指示的产品对象的缺陷检测结果;
其中,所述图像选取方式为基于所述接近传感器与第一相机之间的第一位置差异以及所述接近传感器与第二相机之间的第二位置差异所设定的选取方式。
S801-S803的具体实现方式在上述实施例中都进行过介绍,故本申请实施例在此不做过多赘述。
示例性的,步骤S803,所述针对每一编码信息,按照预定的图像选取方式,选取包含该编码信息所指示的产品对象的第一图像和第二图像,可以包括步骤A1:
A1,针对每一编码信息,基于该编码信息对应的次序、各个第一图像对应的次序以及所述第一位置差异,确定包含该编码信息所指示的产品对象的第一图像;以及,基于该编码信息对应的次序、各个第二图像对应的次序以及所述第二位置差异,确定包含该编码信息所指示的产品对象的第二图像;
其中,该编码信息对应的次序为按照生成时间顺序对各个编码信息进行排序时该编码信息对应的次序;每一第一图像对应的次序为按照拍摄时间对各个第一图像进行排序时,该第一图像对应的次序;每一第二图像对应的次序为按照拍摄时间对各个第二图像进行排序时,该第二图像对应的次序;
其中,所述第一位置差异为:当第一相机的拍摄视野内进入待拍摄的产品对象且存在产品对象接近接近传感器时,进入第一相机的拍摄视野的产品对象与接近所述接近传感器的对象之间,间隔N1个产品对象。
所述第二位置差异为:当第二相机的拍摄视野内进入待拍摄的产品对象且存在产品对象接近接近传感器时,进入第二相机的拍摄视野的产品对象与接近所述接近传感器的对象之间,间隔N2个产品对象;其中,N1和N2为不同的数值。
步骤A1,所述基于该编码信息对应的次序、各个第一图像对应的次序以及所述第一位置差异,确定包含该编码信息所指示的产品对象的第一图像,可以包括步骤B1:
B1:从各个第一图像中,选取所对应次序不小于该编码信息对应的次序且相差N1+1的第一图像。
步骤A1,所述基于该编码信息对应的次序、各个第二图像对应的次序以及所述第二位置差异,确定包含该编码信息所指示的产品对象的第二图像,可以包括步骤B2:
B2:从各个第二图像中,选取所对应次序不小于该编码信息对应的次序且相差N2+1的第二图像。
示例性的,在一种实施方式中,所述方法还包括:从第一相机拍摄得到的首张第一图像开始,对各个第一图像按照拍摄时间进行次序编码,得到各个第一图像的编码值;以及从第二相机拍摄得到的首张第二图像开始,对各个第二图像按照拍摄时间进行次序编码,得到各个第二图像的编码值;
步骤B1,所述从各个第一图像中,选取所对应次序不小于该编码信息对应的次序且相差N1+1的第一图像,可以包括步骤B11:
B11:从各张第一图像中,确定所具有编码值符合第一条件的第一图像,得到所对应次序不小于该编码信息对应的次序且相差N1+1的第一图像;其中,所述第一条件为所具
有编码值不小于该编码信息对应的次序且相差N1+1。
步骤B2,所述从各个第二图像中,选取所对应次序不小于该编码信息对应的次序且相差N2+1的第二图像,可以包括步骤B12:
B12:从各张第二图像中,确定所具有编码值符合第二条件的第二图像,得到所对应次序不小于该编码信息对应的次序且相差N2+1的第二图像;其中,所述第二条件为所具有编码值不小于该编码信息对应的次序且相差N2+1。
示例性的,在另一种实施方式中,所述方法还可以包括:从第一相机拍摄得到的第N1+1张第一图像的下一图像开始,对各个第一图像按照拍摄时间进行次序编码,得到各个第一图像的编码值;以及从第二相机拍摄得到的第N2+1张第二图像的下一图像开始,对各个第二图像按照拍摄时间进行次序编码,得到各个第二图像的编码值;
步骤B1,所述从各个第一图像中,选取所对应次序不小于该编码信息对应的次序且相差N1+1的第一图像,还可以包括步骤B21:
B21,从各个第一图像中,确定所具有编码值与该编码信息对应的次序相匹配的第一图像,得到所对应次序不小于该编码信息对应的次序且相差N1+1的第一图像。
步骤B2,所述从各个第二图像中,选取所对应次序不小于该编码信息对应的次序且相差N2+1的第二图像,还可以包括步骤B22:
B22,从各个第二图像中,确定所具有编码值与该编码信息对应的次序相匹配的第二图像,得到所对应次序不小于该编码信息对应的次序且相差N2+1的第二图像。
示例性的,在一种实施方式中,应用于缺陷检测***的控制装置的缺陷检测方法,还可以包括步骤C1:
在检测到任一编码信息所指示的产品对象的缺陷检测结果,符合剔除条件时,控制所述剔除装置从流水线上剔除该编码信息所指示的产品对象。
步骤C1,所述控制所述剔除装置从流水线上剔除该编码信息所指示的产品对象,还可以包括步骤C11:
C11,在接收到指定感应信号之后,第N3+1次接收到感应信号时,向所述剔除装置发送对象剔除指令,以使所述剔除装置从流水线上剔除该编码信息所指示的产品对象。
其中,所述指定感应信号为该编码信息所表示的产品对象接近时所发出的感应信号。
示例性的,在一种实施方式中,应用于缺陷检测***的控制装置的缺陷检测方法,还可以包括如下步骤:
D1,接收各个第三图像和各个第四图像。
D2,针对每一编码信息,确定与所选取的第一图像的拍摄时间相匹配的第三图像,以及与所选取的第二图像的拍摄时间相匹配的第四图像,对所选取的第三图像和第四图像进行产品模具缺陷检测处理,得到模具检测结果,将所得到的模具检测结果与该编码信息所指示的产品对象的缺陷检测结果进行关联。
上述方法步骤的具体实现方式在其他实施例中已做描述,本实施例中不做过多赘述。
综上所述,本申请实施例中,光检装置获取到的流水线上产品对象的图像是包括了产品对象多面的图像,并且,控制装置可以基于预定的图像选取方式确定出关于每一产品对象对应的图像,即包含任一编码信息所指示的产品对象的第一图像和第二图像,进而对关
于每一产品对象的图像进行产品对象缺陷检测处理,基于产品对象缺陷检测处理得到的结果,可以确定对应产品对象的关于双面的缺陷检测结果。可见,相对于现有技术采集单面图像进行缺陷检测处理而言,本方案的缺陷检测***基于多面视觉缺陷检测以及对各面对应的缺陷检测结果进行关联判定,来确定产品对象的缺陷检测结果,这样可以不存在视野盲区,缺陷检测结果的完整性大大提升,因此,可以提高产品对象的缺陷检测结果的准确率。
另外,本方案中,光检装置设置有错位排布的第一相机和第二相机,以对产品对象的多面进行图像采集。可见,通过该种光检装置可以在保证在线全检测的前提下,通过紧凑的设计空间,实现同时对产品对象的多面的图像采集,相比较传统的每一面单独检测的视觉检测***,大大提高了检测效率。
并且,本方案中,基于接近传感器,实现对于产品对象的编码,可以为同一产品对象的多面检测的关联提供实现条件。
为了进一步理解缺陷检测***的工作原理,下面以流水线中各部件按流程实现缺陷检测的步骤,介绍缺陷检测过程。如图9所示,从流水线中各部件流程的角度而言,缺陷检测过程可以包括如下步骤:
S901,每当产品对象接近时,接近传感器向PLC发送感应信号。
S902,PLC每当接收感应信号时,针对当前接近的产品对象进行编码,并将编码信息发送至主控装置,以及向光检装置发送拍照触发信号。
步骤S902之后同时执行步骤S903和S906。
S903,光检装置响应于接收到的拍照触发信号,控制第一相机和第二相机拍照,得到第一图像和第二图像,以及上报所述第一图像和所述第二图像至主控装置。
S904,主控装置接收各个第一图像和各个第二图像。
S905,主控装置针对每一编码信息,按照预定的图像选取方式,选取包含该编码信息所指示的产品对象的第一图像和第二图像,并对所选取的第一图像以及第二图像进行产品对象缺陷检测处理,基于所述产品对象缺陷检测处理所得到的结果,确定该编码信息所指示的产品对象的缺陷检测结果。
S906,光检装置响应于接收到的拍照触发信号,控制第三相机和第四相机拍照,得到第三图像和第四图像,以及上报所述第三图像和所述第四图像至主控装置。步骤S906之后执行步骤S907。
S907,主控装置接收各个第三图像和各个第四图像。
S908,主控装置针对每一编码信息,确定与所选取的第一图像的拍摄时间相匹配的第三图像,以及与所选取的第二图像的拍摄时间相匹配的第四图像,对所选取的第三图像和第四图像进行产品模具缺陷检测处理,基于所述产品模具缺陷检测处理所得到的结果,确定模具检测结果。
步骤S905和步骤S908之后都执行步骤S909。
S909,主控装置将所得到的模具检测结果与该编码信息所指示的产品对象的缺陷检测结果进行关联。
S9010,主控装置判断是否需要从流水线剔除该编码信号所指示的产品对象。如果是,执行步骤S9011;如果否,执行步骤S9013;
示例性的,主控装置可以按照预定的剔除规则判断产品对象是否需要剔除。
S9011,主控装置向剔除装置发送剔除指令。
示例性的,主控装置可以在判断产品对象需要剔除后,向剔除装置发送剔除指令;
S9012,剔除装置接收主控装置发出的剔除指令,并执行剔除操作。
示例性的,剔除装置可以接收主控装置发出的剔除指令,并采用剔除装置预定的剔除手段执行剔除操作。
S9013,主控装置不向剔除装置发送剔除指令。
S901-S909的具体实现方式在上述实施例中都进行过介绍,故本申请实施例在此不做过多赘述。
综上所述,本申请实施例中,光检装置获取到的流水线上产品对象的图像是包括了产品对象多面的图像,并且,控制装置可以基于预定的图像选取方式确定出关于每一产品对象对应的图像,即包含任一编码信息所指示的产品对象的第一图像和第二图像,进而对关于每一产品对象的图像进行产品对象缺陷检测处理,基于产品对象缺陷检测处理得到的结果,可以确定对应产品对象的关于双面的缺陷检测结果。可见,相对于现有技术采集单面图像进行缺陷检测处理而言,本方案的缺陷检测***基于多面视觉缺陷检测以及对各面对应的缺陷检测结果进行关联判定,来确定产品对象的缺陷检测结果,这样可以不存在视野盲区,缺陷检测结果的完整性大大提升,因此,可以提高产品对象的缺陷检测结果的准确率。
另外,本方案中,光检装置设置有错位排布的第一相机和第二相机,以对产品对象的多面进行图像采集。可见,通过该种光检装置可以在保证在线全检测的前提下,通过紧凑的设计空间,实现同时对产品对象的多面的图像采集,相比较传统的每一面单独检测的视觉检测***,大大提高了检测效率。
并且,本方案中,基于接近传感器,实现对于产品对象的编码,可以为同一产品对象的多面检测的关联提供实现条件。
基于上述的缺陷检测***,从控制装置的角度,本申请实施例提供了一种缺陷检测装置。如图10所示,该缺陷检测装置应用于缺陷检测***的控制装置,可以包括如下模块:
发送模块1010,设置为每当接收到接近传感器发送的感应信号时,针对当前接近的产品对象进行编码,得到当前接近的产品对象的编码信息,以及向所述光检装置发送拍照触发信号,以使光检装置响应于接收到的拍照触发信号,控制所述第一相机和第二相机进行拍照,得到第一图像和第二图像,并上报所述第一图像和第二图像至所述控制装置;其中,所述接近传感器每当产品对象接近时向控制装置发送感应信号;
接收模块1020,设置为接收各个第一图像和各个第二图像;
检测模块1030,设置为针对每一编码信息,按照预定的图像选取方式,选取包含该编码信息所指示的产品对象的第一图像和第二图像,并对所选取的第一图像以及第二图像进行产品对象缺陷检测处理,得到该编码信息所指示的产品对象的缺陷检测结果;其中,所
述图像选取方式为基于所述接近传感器与第一相机之间的第一位置差异以及所述接近传感器与第二相机之间的第二位置差异所设定的选取方式。
上述装置的具体实现方式在其他实施例中已做描述,本实施例中不做过多赘述。
综上所述,本申请实施例中,光检装置获取到的流水线上产品对象的图像是包括了产品对象多面的图像,并且,控制装置可以基于预定的图像选取方式确定出关于每一产品对象对应的图像,即包含任一编码信息所指示的产品对象的第一图像和第二图像,进而对关于每一产品对象的图像进行产品对象缺陷检测处理,基于产品对象缺陷检测处理得到的结果,可以确定对应产品对象的关于双面的缺陷检测结果。可见,相对于现有技术采集单面图像进行缺陷检测处理而言,本方案的缺陷检测***基于多面视觉缺陷检测以及对各面对应的缺陷检测结果进行关联判定,来确定产品对象的缺陷检测结果,这样可以不存在视野盲区,缺陷检测结果的完整性大大提升,因此,可以提高产品对象的缺陷检测结果的准确率。
另外,本方案中,光检装置设置有错位排布的第一相机和第二相机,以对产品对象的多面进行图像采集。可见,通过该种光检装置可以在保证在线全检测的前提下,通过紧凑的设计空间,实现同时对产品对象的多面的图像采集,相比较传统的每一面单独检测的视觉检测***,大大提高了检测效率。
并且,本方案中,基于接近传感器,实现对于产品对象的编码,可以为同一产品对象的多面检测的关联提供实现条件。
本申请实施例还提供了一种电子设备,如图11所示,包括处理器1101、通信接口1102、存储器1103和通信总线1104,其中,处理器1101,通信接口1102,存储器1103通过通信总线1104完成相互间的通信,
存储器1103,用于存放计算机程序;
处理器1101,用于执行存储器1103上所存放的程序时,实现上述缺陷检测方法。
上述电子设备提到的通信总线可以是外设部件互连标准(Peripheral Component Interconnect,PCI)总线或扩展工业标准结构(Extended Industry Standard Architecture,EISA)总线等。该通信总线可以分为地址总线、数据总线、控制总线等。为便于表示,图中仅用一条粗线表示,但并不表示仅有一根总线或一种类型的总线。
通信接口用于上述电子设备与其他设备之间的通信。
存储器可以包括随机存取存储器(Random Access Memory,RAM),也可以包括非易失性存储器(Non-Volatile Memory,NVM),例如至少一个磁盘存储器。可选的,存储器还可以是至少一个位于远离前述处理器的存储装置。
上述的处理器可以是通用处理器,包括中央处理器(Central Processing Unit,CPU)、网络处理器(Network Processor,NP)等;还可以是数字信号处理器(Digital Signal Processor,DSP)、专用集成电路(Application Specific Integrated Circuit,ASIC)、现场可编程门阵列(Field-Programmable Gate Array,FPGA)或者其他可编程逻辑器件、分立门或者晶体管逻辑器件、分立硬件组件。
在本申请提供的又一实施例中,还提供了一种计算机可读存储介质,该计算机可读存
储介质内存储有计算机程序,所述计算机程序被处理器执行时实现上述任一缺陷检测方法的步骤。
在本申请提供的又一实施例中,还提供了一种包含指令的计算机程序产品,当其在计算机上运行时,使得计算机执行上述实施例中任一缺陷检测方法。
在本申请提供的又一实施例中,还提供了一种包含指令的计算机程序,所述计算机程序在计算机上运行时,使得计算机执行上述实施例中任一缺陷检测方法。
在上述实施例中,可以全部或部分地通过软件、硬件、固件或者其任意组合来实现。当使用软件实现时,可以全部或部分地以计算机程序产品的形式实现。所述计算机程序产品包括一个或多个计算机指令。在计算机上加载和执行所述计算机程序指令时,全部或部分地产生按照本申请实施例所述的流程或功能。所述计算机可以是通用计算机、专用计算机、计算机网络、或者其他可编程装置。所述计算机指令可以存储在计算机可读存储介质中,或者从一个计算机可读存储介质向另一个计算机可读存储介质传输,例如,所述计算机指令可以从一个网站站点、计算机、服务器或数据中心通过有线(例如同轴电缆、光纤、数字用户线(DSL))或无线(例如红外、无线、微波等)方式向另一个网站站点、计算机、服务器或数据中心进行传输。所述计算机可读存储介质可以是计算机能够存取的任何可用介质或者是包含一个或多个可用介质集成的服务器、数据中心等数据存储设备。所述可用介质可以是磁性介质,(例如,软盘、硬盘、磁带)、光介质(例如,DVD)、或者其他介质(例如固态硬盘Solid State Disk(SSD))等。
需要说明的是,在本文中,诸如第一和第二等之类的关系术语仅仅用来将一个实体或者操作与另一个实体或操作区分开来,而不一定要求或者暗示这些实体或操作之间存在任何这种实际的关系或者顺序。而且,术语“包括”、“包含”或者其任何其他变体意在涵盖非排他性的包含,从而使得包括一系列要素的过程、方法、物品或者设备不仅包括那些要素,而且还包括没有明确列出的其他要素,或者是还包括为这种过程、方法、物品或者设备所固有的要素。在没有更多限制的情况下,由语句“包括一个……”限定的要素,并不排除在包括所述要素的过程、方法、物品或者设备中还存在另外的相同要素。
本说明书中的各个实施例均采用相关的方式描述,各个实施例之间相同相似的部分互相参见即可,每个实施例重点说明的都是与其他实施例的不同之处。尤其,对于方法实施例而言,由于其基本相似于***实施例,所以描述的比较简单,相关之处参见***实施例的部分说明即可。
以上所述仅为本申请的较佳实施例,并非用于限定本申请的保护范围。凡在本申请的精神和原则之内所作的任何修改、等同替换、改进等,均包含在本申请的保护范围内。
Claims (16)
- 一种缺陷检测***,包括:控制装置、光检装置和接近传感器;其中,所述光检装置设置有错位排布的第一相机和第二相机,所述第一相机为对流水线传递的待检测的每一产品对象的一面进行拍照的相机,所述第二相机为对每一产品对象的另一面进行拍照的相机;所述接近传感器,设置在所述流水线上,设置为每当产品对象接近时,向所述控制装置发送感应信号;所述控制装置,设置为每当接收到所述感应信号时,针对当前接近的产品对象进行编码,得到当前接近的产品对象的编码信息,以及向所述光检装置发送拍照触发信号;所述光检装置,设置为响应于接收到的拍照触发信号,控制所述第一相机和第二相机进行拍照,得到第一图像和第二图像,并上报所述第一图像和第二图像至所述控制装置;所述控制装置,还设置为接收各个第一图像和各个第二图像;以及,针对每一编码信息,按照预定的图像选取方式,选取包含该编码信息所指示的产品对象的第一图像和第二图像,并对所选取的第一图像以及第二图像进行产品对象缺陷检测处理,基于所述产品对象缺陷检测处理所得到的结果,确定该编码信息所指示的产品对象的缺陷检测结果;其中,所述图像选取方式为基于所述接近传感器与第一相机之间的第一位置差异以及所述接近传感器与第二相机之间的第二位置差异所设定的选取方式。
- 根据权利要求1所述的***,其中,所述控制装置针对每一编码信息,按照预定的图像选取方式,选取包含该编码信息所指示的产品对象的第一图像和第二图像,包括:针对每一编码信息,基于该编码信息对应的次序、各个第一图像对应的次序以及所述第一位置差异,确定包含该编码信息所指示的产品对象的第一图像;以及,基于该编码信息对应的次序、各个第二图像对应的次序以及所述第二位置差异,确定包含该编码信息所指示的产品对象的第二图像;其中,该编码信息对应的次序为按照生成时间顺序对各个编码信息进行排序时该编码信息对应的次序;每一第一图像对应的次序为按照拍摄时间对各个第一图像进行排序时,该第一图像对应的次序;每一第二图像对应的次序为按照拍摄时间对各个第二图像进行排序时,该第二图像对应的次序。
- 根据权利要求2所述的***,其中,所述第一位置差异为:当第一相机的拍摄视野内进入待拍摄的产品对象且存在产品对象接近接近传感器时,进入第一相机的拍摄视野的产品对象与接近所述接近传感器的产品对象之间,间隔N1个产品对象;所述第二位置差异为:当第二相机的拍摄视野内进入待拍摄的产品对象且存在产品对象接近接近传感器时,进入第二相机的拍摄视野的产品对象与接近所述接近传感器的产品对象之间,间隔N2个产品对象;其中,N1和N2为不同的数值;所述控制装置基于该编码信息对应的次序、各个第一图像对应的次序以及所述第一位置差异,确定包含该编码信息所指示的产品对象的第一图像,包括:从各个第一图像中,选取所对应次序不小于该编码信息对应的次序且相差N1+1的第一图像;所述控制装置基于该编码信息对应的次序、各个第二图像对应的次序以及所述第二位 置差异,确定包含该编码信息所指示的产品对象的第二图像,包括:从各个第二图像中,选取所对应次序不小于该编码信息对应的次序且相差N2+1的第二图像。
- 根据权利要求3所述的***,其中,所述控制装置还设置为:从第一相机拍摄得到的首张第一图像开始,对各个第一图像按照拍摄时间进行次序编码,得到各个第一图像的编码值;以及从第二相机拍摄得到的首张第二图像开始,对各个第二图像按照拍摄时间进行次序编码,得到各个第二图像的编码值;所述控制装置从各个第一图像中,选取所对应次序不小于该编码信息对应的次序且相差N1+1的第一图像,包括:从各张第一图像中,确定所具有编码值符合第一条件的第一图像,得到所对应次序不小于该编码信息对应的次序且相差N1+1的第一图像;其中,所述第一条件为所具有编码值不小于该编码信息对应的次序且相差N1+1;所述控制装置从各个第二图像中,选取所对应次序不小于该编码信息对应的次序且相差N2+1的第二图像,包括:从各张第二图像中,确定所具有编码值符合第二条件的第二图像,得到所对应次序不小于该编码信息对应的次序且相差N2+1的第二图像;其中,所述第二条件为所具有编码值不小于该编码信息对应的次序且相差N2+1。
- 根据权利要求3所述的***,其中,所述控制装置还设置为:从第一相机拍摄得到的第N1+1张第一图像的下一图像开始,对各个第一图像按照拍摄时间进行次序编码,得到各个第一图像的编码值;以及从第二相机拍摄得到的第N2+1张第二图像的下一图像开始,对各个第二图像按照拍摄时间进行次序编码,得到各个第二图像的编码值;所述控制装置从各个第一图像中,选取所对应次序不小于该编码信息对应的次序且相差N1+1的第一图像,包括:从各个第一图像中,确定所具有编码值与该编码信息对应的次序相匹配的第一图像,得到所对应次序不小于该编码信息对应的次序且相差N1+1的第一图像;所述控制装置从各个第二图像中,选取所对应次序不小于该编码信息对应的次序且相差N2+1的第二图像,包括:从各个第二图像中,确定所具有编码值与该编码信息对应的次序相匹配的第二图像,得到所对应次序不小于该编码信息对应的次序且相差N2+1的第二图像。
- 根据权利要求1-5任一项所述的***,其中,所述***还包括:剔除装置,其中,所述剔除装置在所述流水线上的位置处于所述光检装置之后;所述控制装置,还设置为在检测到任一编码信息所指示的产品对象的缺陷检测结果,符合剔除条件时,控制所述剔除装置从流水线上剔除该编码信息所指示的产品对象;所述剔除装置,设置为在所述控制装置的控制下,从流水线上剔除该编码信息所表示的对象。
- 根据权利要求6所述的***,其中,所述接近传感器与所述剔除装置的位置差异为:当存在产品对象进入所述剔除装置的可操作区域且存在产品对象接近所述接近传感器时,进入所述可操作区域的产品对象与接近所述接近传感器的产品对象之间,间隔N3个 产品对象;所述控制装置控制所述剔除装置从流水线上剔除该编码信息所指示的产品对象,包括:在接收到指定感应信号之后,第N3+1次接收到感应信号时,向所述剔除装置发送对象剔除指令,以使所述剔除装置从流水线上剔除该编码信息所指示的产品对象;其中,所述指定感应信号为该编码信息所表示的产品对象接近时所发出的感应信号。
- 根据权利要求1-5任一项所述的***,其中,每一产品对象在流水线上传递时,关联有一产品模具;所述光检装置还设置有错位排布的第三相机和第四相机,所述第三相机为对流水线传递的各个产品模具的一面进行拍照的相机,所述第四相机为对各个产品模具的另一面进行拍照的相机;所述光检装置,还设置为响应于接收到的拍照触发信号,控制所述第三相机和第四相机进行拍照,得到第三图像和第四图像,并向所述控制装置上报所述第三图像和第四图像;所述控制装置,还设置为接收各个第三图像和各个第四图像;以及,针对每一编码信息,确定与所选取的第一图像的拍摄时间相匹配的第三图像,以及与所选取的第二图像的拍摄时间相匹配的第四图像,对所选取的第三图像和第四图像进行产品模具缺陷检测处理,基于所述产品模具缺陷检测处理所得到的结果,确定模具检测结果,将所得到的模具检测结果与该编码信息所指示的产品对象的缺陷检测结果进行关联。
- 根据权利要求1-5任一项所述的***,其中,所述流水线的设置为传递产品对象的支架上还设置有限位器。
- 根据权利要求1-5任一项所述的***,其中,所述控制装置包括:主控装置和可编程逻辑控制器PLC;所述PLC,设置为每当接收到所述感应信号时,针对当前接近的产品对象进行编码,并将编码信息发送至主控装置,以及向所述光检装置发送拍照触发信号;所述主控装置,设置为接收各个第一图像和各个第二图像;以及,针对每一编码信息,按照预定的图像选取方式,选取包含该编码信息所指示的产品对象的第一图像和第二图像,并对所选取的第一图像以及第二图像进行产品对象缺陷检测处理,基于所述产品对象缺陷检测处理所得到的结果,确定该编码信息所指示的产品对象的缺陷检测结果;所述光检装置设置为上报所述第一图像和第二图像至所述主控装置。
- 一种基于权利要求1-10任一项所述的缺陷检测***的缺陷检测方法,应用于控制装置;所述方法包括:每当接收到接近传感器发送的感应信号时,针对当前接近的产品对象进行编码,得到当前接近的产品对象的编码信息,以及向所述光检装置发送拍照触发信号,以使光检装置响应于接收到的拍照触发信号,控制所述第一相机和第二相机进行拍照,得到第一图像和第二图像,并上报所述第一图像和第二图像至所述控制装置;其中,所述接近传感器每当产品对象接近时向控制装置发送感应信号;接收各个第一图像和各个第二图像;针对每一编码信息,按照预定的图像选取方式,选取包含该编码信息所指示的产品对象的第一图像和第二图像,并对所选取的第一图像以及第二图像进行产品对象缺陷检测处理,基于所述产品对象缺陷检测处理所得到的结果,确定该编码信息所指示的产品对象的 缺陷检测结果;其中,所述图像选取方式为基于所述接近传感器与第一相机之间的第一位置差异以及所述接近传感器与第二相机之间的第二位置差异所设定的选取方式。
- 一种基于权利要求1-10任一项所述的缺陷检测***的缺陷检测装置,应用于控制装置;所述缺陷检测装置包括:发送模块,设置为每当接收到接近传感器发送的感应信号时,针对当前接近的产品对象进行编码,得到当前接近的产品对象的编码信息,以及向所述光检装置发送拍照触发信号,以使光检装置响应于接收到的拍照触发信号,控制所述第一相机和第二相机进行拍照,得到第一图像和第二图像,并上报所述第一图像和第二图像至所述控制装置;其中,所述接近传感器每当产品对象接近时向控制装置发送感应信号;接收模块,设置为接收各个第一图像和各个第二图像;检测模块,设置为针对每一编码信息,按照预定的图像选取方式,选取包含该编码信息所指示的产品对象的第一图像和第二图像,并对所选取的第一图像以及第二图像进行产品对象缺陷检测处理,基于所述产品对象缺陷检测处理所得到的结果,确定该编码信息所指示的产品对象的缺陷检测结果;其中,所述图像选取方式为基于所述接近传感器与第一相机之间的第一位置差异以及所述接近传感器与第二相机之间的第二位置差异所设定的选取方式。
- 一种电子设备,其中,包括处理器、通信接口、存储器和通信总线,其中,处理器,通信接口,存储器通过通信总线完成相互间的通信;存储器,用于存放计算机程序;处理器,用于执行存储器上所存放的程序时,实现权利要求11所述方法的步骤。
- 一种计算机可读存储介质,其中,所述计算机可读存储介质内存储有计算机程序,所述计算机程序被处理器执行时实现权利要求11所述方法的步骤。
- 一种包含指令的计算机程序产品,所述包含指令的计算机程序产品在计算机上运行时,使得计算机执行权利要求11所述方法的步骤。
- 一种计算机程序,所述计算机程序在计算机上运行时,使得计算机执行权利要求11所述方法的步骤。
Applications Claiming Priority (2)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202211116759.7A CN115423785A (zh) | 2022-09-14 | 2022-09-14 | 一种缺陷检测***、方法、装置、电子设备及存储介质 |
CN202211116759.7 | 2022-09-14 |
Publications (2)
Publication Number | Publication Date |
---|---|
WO2024055796A1 true WO2024055796A1 (zh) | 2024-03-21 |
WO2024055796A9 WO2024055796A9 (zh) | 2024-04-18 |
Family
ID=84202032
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
PCT/CN2023/113105 WO2024055796A1 (zh) | 2022-09-14 | 2023-08-15 | 一种缺陷检测***、方法、装置、电子设备及存储介质 |
Country Status (2)
Country | Link |
---|---|
CN (1) | CN115423785A (zh) |
WO (1) | WO2024055796A1 (zh) |
Families Citing this family (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN115423785A (zh) * | 2022-09-14 | 2022-12-02 | 杭州海康机器人股份有限公司 | 一种缺陷检测***、方法、装置、电子设备及存储介质 |
CN117911411B (zh) * | 2024-03-19 | 2024-05-24 | 南京认知物联网研究院有限公司 | 一种基于图片流并行检测的计算机视觉检测方法及装置 |
Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
WO2012125526A1 (en) * | 2011-03-11 | 2012-09-20 | Accusentry, Inc. | Single culling mechanism for use with machine vison on production line |
CN110487807A (zh) * | 2019-09-20 | 2019-11-22 | 广州市讯思视控科技有限公司 | 一种干电池锌筒外观缺陷视觉检测装置及方法 |
CN111999308A (zh) * | 2020-08-27 | 2020-11-27 | 银河水滴科技(北京)有限公司 | 一种缺陷检测的控制方法、控制装置及控制*** |
CN113436180A (zh) * | 2021-07-07 | 2021-09-24 | 京东科技控股股份有限公司 | 生产线上喷射码的检测方法、装置、***、设备和介质 |
CN113820326A (zh) * | 2021-11-19 | 2021-12-21 | 佛山市煜丰机械有限公司 | 一种长码拉链的缺陷检测*** |
CN115423785A (zh) * | 2022-09-14 | 2022-12-02 | 杭州海康机器人股份有限公司 | 一种缺陷检测***、方法、装置、电子设备及存储介质 |
-
2022
- 2022-09-14 CN CN202211116759.7A patent/CN115423785A/zh active Pending
-
2023
- 2023-08-15 WO PCT/CN2023/113105 patent/WO2024055796A1/zh unknown
Patent Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
WO2012125526A1 (en) * | 2011-03-11 | 2012-09-20 | Accusentry, Inc. | Single culling mechanism for use with machine vison on production line |
CN110487807A (zh) * | 2019-09-20 | 2019-11-22 | 广州市讯思视控科技有限公司 | 一种干电池锌筒外观缺陷视觉检测装置及方法 |
CN111999308A (zh) * | 2020-08-27 | 2020-11-27 | 银河水滴科技(北京)有限公司 | 一种缺陷检测的控制方法、控制装置及控制*** |
CN113436180A (zh) * | 2021-07-07 | 2021-09-24 | 京东科技控股股份有限公司 | 生产线上喷射码的检测方法、装置、***、设备和介质 |
CN113820326A (zh) * | 2021-11-19 | 2021-12-21 | 佛山市煜丰机械有限公司 | 一种长码拉链的缺陷检测*** |
CN115423785A (zh) * | 2022-09-14 | 2022-12-02 | 杭州海康机器人股份有限公司 | 一种缺陷检测***、方法、装置、电子设备及存储介质 |
Also Published As
Publication number | Publication date |
---|---|
CN115423785A (zh) | 2022-12-02 |
WO2024055796A9 (zh) | 2024-04-18 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
WO2024055796A1 (zh) | 一种缺陷检测***、方法、装置、电子设备及存储介质 | |
JP5043430B2 (ja) | 自動ウェブ検査のための装置および方法 | |
CN101696945B (zh) | 机器视觉***对光伏玻璃瑕疵的在线检测方法 | |
CN109946303A (zh) | 检查装置和方法 | |
CN103072718B (zh) | 用于食品包装生产线的机器视觉检测***及方法 | |
CN110490842B (zh) | 一种基于深度学习的带钢表面缺陷检测方法 | |
KR20190122550A (ko) | 검사 관리 시스템, 검사 관리 장치 및 검사 관리 방법 | |
KR20210008352A (ko) | 촬상된 품목의 결함을 검출하기 위한 시스템 및 방법 | |
CN109462999B (zh) | 通过数据平衡基于学习的视觉检查方法以及利用其的视觉检查装置 | |
CN111310826B (zh) | 样本集的标注异常检测方法、装置及电子设备 | |
TWI747686B (zh) | 缺陷檢測方法及檢測裝置 | |
TW202127014A (zh) | 智能產線監測系統及監測方法 | |
CN102303018B (zh) | 机器视觉***对注射器钢针毛刺的自动检测方法 | |
TW202242390A (zh) | 缺陷檢查裝置、缺陷檢查方法以及製造方法 | |
Wang et al. | Design of machine vision applications in detection of defects in high-speed bar copper | |
TWI694250B (zh) | 表面缺陷偵測系統及其方法 | |
CN216525503U (zh) | 一种基于机器视觉的碳纤维预浸料表面缺陷在线检测装置 | |
CN115218790A (zh) | 棒材检测方法、装置和*** | |
CN114226262A (zh) | 瑕疵检测方法、瑕疵分类方法及其*** | |
TWI749714B (zh) | 瑕疵檢測方法、瑕疵分類方法及其系統 | |
CN105973910A (zh) | 基于结构纹理特征的灯管质量检测与瑕疵分类方法及*** | |
CN102305603A (zh) | 机器视觉***对注射器钢针倾斜的自动检测方法 | |
TWM606740U (zh) | 瑕疵檢測系統 | |
JP2011232302A (ja) | 画像検査方法及び画像検査装置 | |
US20210049396A1 (en) | Optical quality control |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
121 | Ep: the epo has been informed by wipo that ep was designated in this application |
Ref document number: 23864530 Country of ref document: EP Kind code of ref document: A1 |