CN114372984A - Super-resolution component angle identification device and method - Google Patents
Super-resolution component angle identification device and method Download PDFInfo
- Publication number
- CN114372984A CN114372984A CN202210279738.0A CN202210279738A CN114372984A CN 114372984 A CN114372984 A CN 114372984A CN 202210279738 A CN202210279738 A CN 202210279738A CN 114372984 A CN114372984 A CN 114372984A
- Authority
- CN
- China
- Prior art keywords
- image
- target component
- component
- resolution
- resolution picture
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 34
- 238000001514 detection method Methods 0.000 claims abstract description 17
- 238000012545 processing Methods 0.000 claims description 43
- 238000013136 deep learning model Methods 0.000 claims description 19
- 238000013528 artificial neural network Methods 0.000 claims description 15
- 238000004891 communication Methods 0.000 claims description 12
- 239000004973 liquid crystal related substance Substances 0.000 claims description 9
- 230000005540 biological transmission Effects 0.000 claims description 8
- 230000000877 morphologic effect Effects 0.000 claims description 6
- 238000007781 pre-processing Methods 0.000 claims description 5
- 238000005549 size reduction Methods 0.000 claims description 2
- 238000012360 testing method Methods 0.000 abstract description 11
- 238000004519 manufacturing process Methods 0.000 abstract description 5
- 238000003672 processing method Methods 0.000 abstract description 2
- 230000009286 beneficial effect Effects 0.000 abstract 1
- 230000006870 function Effects 0.000 description 9
- 230000008569 process Effects 0.000 description 7
- 238000012549 training Methods 0.000 description 6
- 238000013507 mapping Methods 0.000 description 4
- 238000010586 diagram Methods 0.000 description 3
- 238000005516 engineering process Methods 0.000 description 3
- 230000005284 excitation Effects 0.000 description 3
- 238000000605 extraction Methods 0.000 description 3
- RSWGJHLUYNHPMX-UHFFFAOYSA-N Abietic-Saeure Natural products C12CCC(C(C)C)=CC2=CCC2C1(C)CCCC2(C)C(O)=O RSWGJHLUYNHPMX-UHFFFAOYSA-N 0.000 description 1
- KHPCPRHQVVSZAH-HUOMCSJISA-N Rosin Natural products O(C/C=C/c1ccccc1)[C@H]1[C@H](O)[C@@H](O)[C@@H](O)[C@@H](CO)O1 KHPCPRHQVVSZAH-HUOMCSJISA-N 0.000 description 1
- 238000007792 addition Methods 0.000 description 1
- 230000003321 amplification Effects 0.000 description 1
- 238000013527 convolutional neural network Methods 0.000 description 1
- 230000007797 corrosion Effects 0.000 description 1
- 238000005260 corrosion Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000003199 nucleic acid amplification method Methods 0.000 description 1
- 238000003860 storage Methods 0.000 description 1
- 230000002194 synthesizing effect Effects 0.000 description 1
- KHPCPRHQVVSZAH-UHFFFAOYSA-N trans-cinnamyl beta-D-glucopyranoside Natural products OC1C(O)C(O)C(CO)OC1OCC=CC1=CC=CC=C1 KHPCPRHQVVSZAH-UHFFFAOYSA-N 0.000 description 1
- 230000000007 visual effect Effects 0.000 description 1
Images
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
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T3/00—Geometric image transformations in the plane of the image
- G06T3/40—Scaling of whole images or parts thereof, e.g. expanding or contracting
- G06T3/4023—Scaling of whole images or parts thereof, e.g. expanding or contracting based on decimating pixels or lines of pixels; based on inserting pixels or lines of pixels
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- General Health & Medical Sciences (AREA)
- General Engineering & Computer Science (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- Artificial Intelligence (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- Biomedical Technology (AREA)
- Life Sciences & Earth Sciences (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Health & Medical Sciences (AREA)
- Quality & Reliability (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Image Analysis (AREA)
- Image Processing (AREA)
Abstract
The invention provides a super-resolution component angle recognition device and method, which are used for collecting an image of a component, amplifying the detail characteristics of a collected target component, accurately calculating to obtain component mounting angle information by combining a straight line fitting method, comparing a component angle calculation value with a preset value, and automatically judging whether the component to be tested meets the test requirements. The invention has the beneficial effects that: for small-packaged components, aiming at the problems that the mounting angle of the components is difficult to accurately judge by AOI equipment and human eyes, the detection precision is poor, and the production and detection are difficult to meet, the mounting precision of the components is improved and the running stability of a circuit board is ensured by adopting an image super-resolution processing method.
Description
Technical Field
The invention belongs to the technical field of image processing, and relates to a super-resolution component angle identification device and method.
Background
In the manufacturing industry field of processing, no matter in surface mounting process or last testing process all need place the angle to components and parts and detect, if the angle is placed to components and parts and demand angle difference is too big, then very easily cause this components and parts rosin joint, influence the stability of whole circuit board. For the detection process, a method of human eye detection is adopted in the traditional scheme, but for the condition that the number of components is large or the size of the components is small, the method is easy to generate mistakes and omissions, and the efficiency is low. In other existing schemes, the function of component detection in AOI equipment is adopted for detection, and compared with the traditional human eye detection, the method has the advantages that the accuracy is improved, the manufacturing cost of the AOI equipment is high, manual participation is needed in the detection process, and complete automation is not achieved. And for small-packaged components, the AOI equipment has large angle identification errors due to the pixel limitation of image acquisition equipment, and is difficult to meet the production and detection requirements, so that the problem of detecting the placement angle of the components automatically, efficiently, accurately and at low cost is required to be solved.
Aiming at the defects of the existing scheme, the invention adopts the image processing technology to carry out super-resolution amplification on the acquired component image and then carry out component angle detection, does not need manual participation, changes the traditional subjective visual discrimination into objective image discrimination, can improve the component detection precision, reduces the omission factor and the labor cost, and improves the production test efficiency.
Disclosure of Invention
The invention aims to provide a device and a method for detecting the angle of a component, which can automatically distinguish a qualified product from an unqualified product by acquiring and inputting an image of the component, obtaining the angle information of the component by adopting image super-resolution processing and recognition technology for the input image, and comparing the angle information with a preset value of the angle of the component to judge whether the component to be detected meets the test requirement.
The technical solution for realizing the purpose of the invention is as follows:
firstly, a super-resolution component angle identification method is disclosed, which comprises the following steps:
step 1: acquiring a high-resolution picture-low-resolution picture sample of the component;
step 2: taking a low-resolution picture as model input and a high-resolution picture as model output, and establishing an image super-resolution neural network deep learning model;
and step 3: collecting an image of a target component;
and 4, step 4: preprocessing the target component image to obtain a low-resolution picture of the target component;
and 5: inputting the low-resolution picture of the target component into the super-resolution neural network deep learning model to obtain a high-resolution picture of the target component;
step 6: binarizing the high-resolution picture of the target component, performing morphological processing, and keeping the edge information of the target component; and fitting an optimal straight line on the edge points of the target component, wherein the included angle between the straight line and the horizontal direction is the angle of the target component.
Preferably, it further comprises step 7: and judging whether the mounting error is smaller than an error allowable value or not based on the angle of the target component, and transmitting a judgment result to the execution module.
Preferably, it further comprises a step 8: and the execution module controls the execution mode of the next link according to the judgment result in the step 7.
Specifically, in step 1, a high-resolution picture sample of the component is obtained first, and the high-resolution picture is converted into a low-resolution picture to obtain a low-resolution picture sample.
Specifically, the image size reduction method includes extracting intermediate pixel points between every 3 pixel points from every 9 adjacent pixel points, finally leaving 4 corner points, and repeating the steps until the image size meets the requirements.
Specifically, in step 3, the execution module uses the conveyor belt to move the circuit board to be tested to the working area of the component image acquisition module, and the camera of the component image acquisition module acquires images of the target component.
Specifically, in step 4, the image is preprocessed: and deleting the background part irrelevant to the target component, eliminating interference factors, and keeping the outline and color information of the target component to obtain a low-resolution image of the target component.
Specifically, in step 6, the method for fitting the optimal straight line through the image contour is to make the image contour pointsiTo a straight linelA distance ofr i When is coming into contact withWhen the sum of the two is the minimum value, the straight linelTo fit an optimal straight line, wherein。
The invention also discloses a super-resolution component angle recognition device, which comprises: the device comprises a component image acquisition module, an operation and processing module, a display module, a communication module and an execution module.
And the component image acquisition module comprises a camera and a matched image transmission cable and is used for acquiring component images and transmitting the component images to the operation and processing module for subsequent processing.
And the operation and processing module comprises 1 or more high-performance neural network processors and is used for operating the image super-resolution processing algorithm and transmitting the algorithm result to the display module, the communication module and the execution module. The image super-resolution processing algorithm comprises the following steps:
step 1: and transmitting the high-resolution picture to an operation and processing module, reducing the size of the picture, and converting the high-resolution picture into a low-resolution picture.
Step 2: and (3) taking the low-resolution picture in the step (1) as model input and the high-resolution picture as model output, and establishing an image super-resolution neural network deep learning model.
And step 3: the execution module moves the circuit board to be tested to a working area of the component image acquisition module by using a conveyor belt, and a camera of the component image acquisition module acquires an image of a target component; and transmitting the image data to an operation and processing module and inputting the image data into a high-performance processor.
And 4, step 4: preprocessing an input target component image, deleting a background part irrelevant to the target component, eliminating interference factors, and keeping the outline and color information of the target component to obtain a low-resolution image of the target component.
And 5: and (4) inputting the low-resolution image obtained in the step (4) into the image super-resolution neural network deep learning model established in the step (2), and outputting a high-resolution picture of the target component reconstructed by the model.
Step 6: and (5) binarizing the high-resolution reconstructed picture of the target component in the step (5) and performing morphological processing, and reserving edge information of the target component. And fitting an optimal straight line on the edge points of the target component, wherein the included angle between the straight line and the horizontal direction is the angle of the target component.
And 7: and comparing the calculated angle of the target component with a preset value of the angle of the component, judging whether the mounting error is smaller than an error allowable value, judging whether the target component is qualified or not according to the error allowable value, and transmitting a judgment result to an execution module.
And 8: the execution module controls the execution mode of the next link according to the judgment result in the step 7: if the product is qualified, transmitting the product to the next link; if not, it is detected separately.
And the display module comprises a liquid crystal display and a transmission cable matched with the liquid crystal display, and the algorithm result obtained from the operation and processing module is displayed on the liquid crystal display in real time through the transmission cable.
And the communication module comprises a wireless network communication unit and can be used for accessing a network and transmitting the detection data obtained from the operation and processing module to the cloud.
And the execution module comprises a conveyor belt and a mechanical arm and is used for transmitting components, executing the subsequent control flow according to the judgment result of the operation and processing module, transmitting the components to the next link if the received result is qualified, and separately detecting the components if the received result is unqualified.
Compared with the prior art, the invention adopting the technical scheme has the following technical effects:
1. the scheme of the invention adopts the image processing technology to carry out image detection, does not need manual participation, improves the detection efficiency and reduces the labor cost.
2. The invention adopts the super-resolution scheme to amplify the components shot by the camera and restore the components into high-resolution pictures, so that small packaged components with angles which are difficult to accurately judge by human eyes can accurately obtain the placement angles through an algorithm, and circuit board faults caused by component surface mounting problems are greatly reduced.
Drawings
Fig. 1 is a block diagram of a super-resolution device angle recognition apparatus according to the present invention.
FIG. 2 is a block diagram of a super-resolution device angle recognition method of the present invention.
FIG. 3 is a block diagram of an image super-resolution neural network deep learning model of the invention.
Detailed Description
For a better understanding of the present invention, reference is made to the following detailed description of the invention in connection with the accompanying drawings. In the following description, while detailed descriptions of existing prior art may obscure the subject matter of the present invention, the descriptions will be omitted herein.
As shown in fig. 1, the super-resolution device angle recognition apparatus 10 includes a device image acquisition module 100, an operation and processing module 101, a display module 102, an execution module 103, and a communication module 104. The invention selects the angle detection of the small package resistor on the circuit board as an embodiment of the invention and is used for explaining the application process of the device.
The component image acquisition module 100 includes a camera 1000 and a matched data transmission cable, and is configured to acquire a small package resistance image, and transmit the acquired image to the operation and processing module 101 for subsequent processing.
The computing and processing module 101 includes a plurality of high performance processors 1010. In one embodiment of the present invention, the operation and processing module 101 receives the target image collected by the camera 1000 for executing the image processing algorithm, and transmits the algorithm result to the display module 102.
The display module 102 includes a liquid crystal display panel 1020. And the liquid crystal display screen 1020 displays the algorithm result sent by the operation and processing module 101 in real time.
The execution module 103 comprises a conveyor belt and a mechanical arm 1030, is used for transmitting the circuit board, executes the subsequent control flow according to the judgment result of the operation and processing module 101, transmits the circuit board to the next link if the received result is qualified, and independently detects the circuit board if the received result is unqualified.
The communication module 104 includes a wireless network communication unit 1040. The wireless network communication unit 1040 is used for network data transmission, and transmits the image generated by the algorithm and the detection result to the database for storage, so as to be used for product tracing.
The execution module 103 uses a conveyor belt and a mechanical arm 1030 to move the circuit board to an acquisition range of the component image acquisition module 100, and the component image acquisition module 100 acquires an image of the component, wherein the component image acquisition module selected in the embodiment of the invention is the camera 1000. The component image acquisition module 100 transmits image data to the operation and processing module 101, and the image data and a prestored preset value of the angle of the component are input into the high-performance processor 1010 together, and the selected processor (preferably RK 3399) has the operation capability of a complex algorithm model and is calculated through an image processing algorithm.
The invention also comprises a super-resolution component angle identification method, which comprises the following two parts: and obtaining a high-resolution picture of the target component and determining the angle of the target component. With reference to fig. 2, the method for identifying the angle of the super-resolution device includes the following steps:
step 1: in the embodiment of the present invention, 80000 prepared high resolution pictures with an image size of 200 × 200 are transmitted to the operation and processing module 101, the image size is reduced by extracting intermediate pixel points between every 3 pixel points from every 9 adjacent pixel points and finally leaving 4 corner points, the above steps are repeated until the image size satisfies 25 × 25, and the high resolution pictures are converted into low resolution pictures. And synthesizing the high-resolution pictures and the corresponding low-resolution pictures to obtain a sample set.
Step 2:
(1) and (3) taking the low-resolution picture with the image size of 25 multiplied by 25 obtained in the step (1) as model input, taking the high-resolution picture with the image size of 200 multiplied by 200 as model output, and establishing an image super-resolution neural network deep learning model. In one embodiment of the present invention, 80% of the sample set obtained in step 1 is used as the training set, and 20% of the sample set is used as the testing set.
(2) And constructing an image super-resolution neural network deep learning model. The deep learning model is composed of a convolutional neural network (as shown in fig. 3), and is divided into three parts, namely a feature extraction module, a feature mapping module and a reconstruction module. The feature extraction module is mainly used for extracting abstract features of the low-resolution image; the characteristic mapping module is mainly used for establishing characteristic mapping from a low-resolution image to a high-resolution image; the reconstruction module is mainly used for reconstructing a high-resolution image. In one embodiment of the invention, the convolution kernel size in feature extraction module convolution layer one is 9 × 9, the number of convolution kernels is 128, and the excitation function is ReLU. The convolution kernel size in the feature mapping module convolution layer two is 1 × 1, the number of convolution kernels is 64, and the excitation function is ReLU. The convolution kernel size in the reconstruction module deconvolution layer I and the deconvolution layer II is 5 multiplied by 5, the number of convolution kernels is 3, and the excitation function is ReLU. The step length of the first deconvolution layer is 4, and the step length of the second deconvolution layer is 2.
(3) Initializing all parameters of the image super-resolution neural network deep learning model, setting model training ending conditions, and completing the training process of the deep learning model by using training samples which account for 80% of a sample set; and (3) in the training process, the low-resolution picture obtained in the step (1) is used as the input of the deep learning model, the high-resolution picture is used as the output of the deep learning model, and all parameters of the deep learning model are continuously adjusted until a training termination condition (namely the maximum iteration number) is reached.
(4) The test set accounting for 20% of the sample set includes low resolution pictures and their corresponding high resolution pictures. And inputting the low-resolution picture of the test sample into the trained deep learning model to obtain the test output corresponding to the test sample, namely the high-resolution reconstructed picture of the test sample. Then, comparing the high-resolution reconstructed picture output by the model with the high-resolution original picture in the test sample set, evaluating the quality of the model by calculating the result size of the loss function, and carrying out model hyper-parameter adjustment to determine a final model function;
the loss function is a mean square loss function:
And step 3: the execution module 103 moves the circuit board to be tested to the working area of the component image acquisition module 100 by using a conveyor belt and a mechanical arm 1030, and the camera 1000 of the component image acquisition module 100 acquires an image of a target component; the image data is transmitted to the arithmetic and processing block 101 and input to the high-performance processor 1010.
And 4, step 4: preprocessing an input target component image, deleting a background part irrelevant to the target component, eliminating interference factors, retaining the outline and color information of the target component, obtaining a low-resolution image of the target component, and unifying the size of the low-resolution image. In one embodiment of the invention, the low resolution image size is 25 × 25.
And 5: and (4) inputting the low-resolution image obtained in the step (4) into the image super-resolution neural network deep learning model established in the step (2), and outputting a high-resolution picture of the target component reconstructed by the model. In one embodiment of the invention, the high resolution image size resulting from the model reconstruction is 200 x 200.
Step 6: and (5) binarizing the high-resolution reconstructed picture of the target component in the step (5) and performing morphological processing, and reserving edge information of the target component. In an embodiment of the present invention, the threshold for the picture binarization is 60. The morphological processing method comprises the steps of firstly performing expansion operation on the binarized picture for 2 times by taking the size of 3 multiplied by 3 as a core size, and then performing corrosion operation for 3 times by taking the size of 3 multiplied by 3 as the core size. And fitting an optimal straight line on the edge points of the target component, wherein the included angle between the straight line and the horizontal direction is the angle of the target component. In one embodiment of the present invention, the method for fitting the image contour to the optimal straight line is to make the image contour pointsiTo a straight linelA distance ofr i When is coming into contact withWhen the sum of the two is the minimum value, the straight linelTo fit an optimal straight line, wherein
And 7: and comparing the calculated target component angle with a component angle preset value, judging whether the mounting error is smaller than an error allowable value, judging whether the target component is qualified or not according to the judgment, and transmitting a judgment result to the execution module 103 and the display module 102. The display module 102 displays the judgment result in real time through the liquid crystal display screen 1020; in one embodiment of the invention, when the deviation of the target component angle and the preset component angle value is more than 0.1 degree, the product is judged to be unqualified.
And 8: the execution module 103 controls the next link execution mode according to the judgment result in the step 7: if the product is qualified, transmitting the product to the next link; if not, the film is detected separately for manual subsequent processing.
The specific embodiments described herein are merely illustrative of the spirit of the invention. Various modifications or additions may be made to the described embodiments or alternatives may be employed by those skilled in the art without departing from the spirit or ambit of the invention as defined in the appended claims.
Claims (9)
1. A super-resolution component angle identification method is characterized by comprising the following steps:
step 1: acquiring a high-resolution picture-low-resolution picture sample of the component;
step 2: taking a low-resolution picture as model input and a high-resolution picture as model output, and establishing an image super-resolution neural network deep learning model;
and step 3: collecting an image of a target component;
and 4, step 4: preprocessing the target component image to obtain a low-resolution picture of the target component;
and 5: inputting the low-resolution picture of the target component into the super-resolution neural network deep learning model to obtain a high-resolution picture of the target component;
step 6: binarizing the high-resolution picture of the target component, performing morphological processing, and keeping the edge information of the target component; and fitting an optimal straight line on the edge points of the target component, wherein the included angle between the straight line and the horizontal direction is the angle of the target component.
2. The method according to claim 1, characterized in that the method further comprises the step 7: and judging whether the mounting error is smaller than an error allowable value or not based on the angle of the target component, and transmitting a judgment result to the execution module.
3. The method according to claim 1, characterized in that the method further comprises the step 8 of: and the execution module controls the execution mode of the next link according to the judgment result in the step 7.
4. The method according to claim 1, wherein in step 1, a high resolution picture sample of the component is obtained first, and the high resolution picture is converted into a low resolution picture to obtain a low resolution picture sample.
5. The method of claim 4, wherein the image size reduction method comprises extracting intermediate pixels between every 3 pixels from every 9 adjacent pixels, and finally leaving 4 corners, and repeating the above steps until the image size meets the requirement.
6. The method according to claim 1, wherein in step 3, the execution module uses the conveyor belt to move the circuit board to be tested to the working area of the component image acquisition module, and the camera of the component image acquisition module acquires an image of the target component.
7. The method according to claim 1, wherein in step 4, the image is pre-processed: and deleting the background part irrelevant to the target component, eliminating interference factors, and keeping the outline and color information of the target component to obtain a low-resolution image of the target component.
8. The method of claim 1, wherein in step 6, the optimal straight line is fitted by the image contour by fitting the image contour with the image contour pointsiTo a straight linelA distance ofr i When is coming into contact withWhen the sum of the two is the minimum value, the straight linelTo fit an optimal straight line, wherein。
9. A super-resolution component angle recognition device is characterized by comprising a component image acquisition module, an operation and processing module, a display module, a communication module and an execution module;
the component image acquisition module comprises a camera and a matched image transmission cable and is used for acquiring a component image and transmitting the component image to the operation and processing module for subsequent processing;
the operation and processing module comprises at least 1 high-performance neural network processor and is used for operating an image super-resolution processing algorithm and transmitting an algorithm result to the display module, the communication module and the execution module;
the image super-resolution processing algorithm comprises the following steps:
step 1: acquiring a high-resolution picture-low-resolution picture sample of the component;
step 2: taking a low-resolution picture as model input and a high-resolution picture as model output, and establishing an image super-resolution neural network deep learning model;
and step 3: collecting an image of a target component;
and 4, step 4: preprocessing the target component image to obtain a low-resolution picture of the target component;
and 5: inputting the low-resolution picture of the target component into the super-resolution neural network deep learning model to obtain a high-resolution picture of the target component;
step 6: binarizing the high-resolution picture of the target component, performing morphological processing, and keeping the edge information of the target component; fitting an optimal straight line on edge points of the target component, wherein an included angle between the straight line and the horizontal direction is an angle of the target component;
and 7: judging whether the mounting error is smaller than an error allowable value or not based on the angle of the target component, and transmitting a judgment result to an execution module;
and 8: the execution module controls the execution mode of the next link according to the judgment result in the step 7;
the display module comprises a liquid crystal display and a transmission cable matched with the liquid crystal display, and the algorithm result obtained from the operation and processing module is displayed on the liquid crystal display in real time through the transmission cable;
the communication module comprises a wireless network communication unit and is used for accessing a network and transmitting the detection data obtained from the operation and processing module to the cloud;
and the execution module comprises a conveyor belt and a mechanical arm and is used for transmitting components, executing the subsequent control flow according to the judgment result of the operation and processing module, transmitting the components to the next link if the received result is qualified, and separately detecting the components if the received result is unqualified.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202210279738.0A CN114372984A (en) | 2022-03-22 | 2022-03-22 | Super-resolution component angle identification device and method |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202210279738.0A CN114372984A (en) | 2022-03-22 | 2022-03-22 | Super-resolution component angle identification device and method |
Publications (1)
Publication Number | Publication Date |
---|---|
CN114372984A true CN114372984A (en) | 2022-04-19 |
Family
ID=81146636
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202210279738.0A Pending CN114372984A (en) | 2022-03-22 | 2022-03-22 | Super-resolution component angle identification device and method |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN114372984A (en) |
Citations (15)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US6064758A (en) * | 1996-11-27 | 2000-05-16 | Daewoo Electronics Co., Ltd. | Mounting coordinate input method and apparatus for surface mount device |
CN101477066A (en) * | 2009-01-09 | 2009-07-08 | 华南理工大学 | Circuit board element mounting/welding quality detection method and system based on super-resolution image reconstruction |
CN103729655A (en) * | 2014-01-22 | 2014-04-16 | 哈尔滨工业大学 | Detection method for sheet element visual positioning |
CN104359402A (en) * | 2014-11-17 | 2015-02-18 | 南京工业大学 | Detection method for rectangular pin element visual positioning |
CN104915963A (en) * | 2015-06-25 | 2015-09-16 | 哈尔滨工业大学 | Detection and positioning method for PLCC component |
CN104933720A (en) * | 2015-06-25 | 2015-09-23 | 哈尔滨工业大学 | SOP element positioning and defect detecting method based on vision |
CN104981105A (en) * | 2015-07-09 | 2015-10-14 | 广东工业大学 | Detecting and error-correcting method capable of rapidly and accurately obtaining element center and deflection angle |
WO2018192251A1 (en) * | 2017-04-18 | 2018-10-25 | 冀月斌 | Method and system for correcting imported production data of surface-mount device |
CN108780570A (en) * | 2016-01-16 | 2018-11-09 | 菲力尔***公司 | Use the system and method for the image super-resolution of iteration collaboration filtering |
CN109118432A (en) * | 2018-09-26 | 2019-01-01 | 福建帝视信息科技有限公司 | A kind of image super-resolution rebuilding method based on Rapid Circulation convolutional network |
CN112508789A (en) * | 2020-12-16 | 2021-03-16 | 广州佳帆计算机有限公司 | Residual error-based patch image enhancement identification method and device |
CN112529883A (en) * | 2020-12-16 | 2021-03-19 | 广州佳帆计算机有限公司 | Patch detection method and device based on image edge recognition |
CN113409234A (en) * | 2020-03-16 | 2021-09-17 | 耐斯泰科技2001有限公司 | Minimum supervision Automatic Inspection (AI) of wafers supported by Convolutional Neural Network (CNN) algorithm |
CN113792725A (en) * | 2021-11-15 | 2021-12-14 | 南京熊猫电子制造有限公司 | Component detection device and method |
CN114004815A (en) * | 2021-11-03 | 2022-02-01 | 哈工大机器人(合肥)国际创新研究院 | PCBA appearance detection method and device |
-
2022
- 2022-03-22 CN CN202210279738.0A patent/CN114372984A/en active Pending
Patent Citations (15)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US6064758A (en) * | 1996-11-27 | 2000-05-16 | Daewoo Electronics Co., Ltd. | Mounting coordinate input method and apparatus for surface mount device |
CN101477066A (en) * | 2009-01-09 | 2009-07-08 | 华南理工大学 | Circuit board element mounting/welding quality detection method and system based on super-resolution image reconstruction |
CN103729655A (en) * | 2014-01-22 | 2014-04-16 | 哈尔滨工业大学 | Detection method for sheet element visual positioning |
CN104359402A (en) * | 2014-11-17 | 2015-02-18 | 南京工业大学 | Detection method for rectangular pin element visual positioning |
CN104915963A (en) * | 2015-06-25 | 2015-09-16 | 哈尔滨工业大学 | Detection and positioning method for PLCC component |
CN104933720A (en) * | 2015-06-25 | 2015-09-23 | 哈尔滨工业大学 | SOP element positioning and defect detecting method based on vision |
CN104981105A (en) * | 2015-07-09 | 2015-10-14 | 广东工业大学 | Detecting and error-correcting method capable of rapidly and accurately obtaining element center and deflection angle |
CN108780570A (en) * | 2016-01-16 | 2018-11-09 | 菲力尔***公司 | Use the system and method for the image super-resolution of iteration collaboration filtering |
WO2018192251A1 (en) * | 2017-04-18 | 2018-10-25 | 冀月斌 | Method and system for correcting imported production data of surface-mount device |
CN109118432A (en) * | 2018-09-26 | 2019-01-01 | 福建帝视信息科技有限公司 | A kind of image super-resolution rebuilding method based on Rapid Circulation convolutional network |
CN113409234A (en) * | 2020-03-16 | 2021-09-17 | 耐斯泰科技2001有限公司 | Minimum supervision Automatic Inspection (AI) of wafers supported by Convolutional Neural Network (CNN) algorithm |
CN112508789A (en) * | 2020-12-16 | 2021-03-16 | 广州佳帆计算机有限公司 | Residual error-based patch image enhancement identification method and device |
CN112529883A (en) * | 2020-12-16 | 2021-03-19 | 广州佳帆计算机有限公司 | Patch detection method and device based on image edge recognition |
CN114004815A (en) * | 2021-11-03 | 2022-02-01 | 哈工大机器人(合肥)国际创新研究院 | PCBA appearance detection method and device |
CN113792725A (en) * | 2021-11-15 | 2021-12-14 | 南京熊猫电子制造有限公司 | Component detection device and method |
Non-Patent Citations (6)
Title |
---|
CHAO DONG 等: "Accelerating the Super-Resolution Convolutional Neural Network", 《ARXIV》 * |
CHAO DONG 等: "Learning a Deep Convolutional Network for Image Super-Resolution", 《ECCV 2014》 * |
HUIJUN GAO 等: "A Line-Based-Clustering Approach for Ball Grid Array Component Inspection in Surface-Mount Technology", 《IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS》 * |
刘瑞祯,于仕琪: "《OpenCV教程-基础篇》", 30 June 2007 * |
杜思思: "电子元器件图像外形特征的精确定位技术研究", 《武汉理工大学学报 信息与管理工程版》 * |
罗振威: "基于机器视觉的贴片机定位算法研究", 《中国优秀硕士学位论文全文数据库 信息科技辑》 * |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN106875381B (en) | Mobile phone shell defect detection method based on deep learning | |
CN108765416B (en) | PCB surface defect detection method and device based on rapid geometric alignment | |
CN109635806B (en) | Ammeter value identification method based on residual error network | |
CN112651968B (en) | Wood board deformation and pit detection method based on depth information | |
WO2023168972A1 (en) | Linear array camera-based copper surface defect detection method and apparatus | |
CN106248686A (en) | Glass surface defects based on machine vision detection device and method | |
CN111598771B (en) | PCB (printed Circuit Board) defect detection system and method based on CCD (Charge coupled device) camera | |
CN109949725B (en) | Image gray level standardization method and system for AOI system | |
CN109872309A (en) | Detection system, method, apparatus and computer readable storage medium | |
CN111415339B (en) | Image defect detection method for complex texture industrial product | |
CN114821376B (en) | Unmanned aerial vehicle image geological disaster automatic extraction method based on deep learning | |
CN111157532A (en) | Visual detection device and method for scratches of mobile phone shell | |
CN113688817A (en) | Instrument identification method and system for automatic inspection | |
CN115830004A (en) | Surface defect detection method, device, computer equipment and storage medium | |
CN108871185B (en) | Method, device and equipment for detecting parts and computer readable storage medium | |
CN116071315A (en) | Product visual defect detection method and system based on machine vision | |
CN115775236A (en) | Surface tiny defect visual detection method and system based on multi-scale feature fusion | |
CN112381751A (en) | Online intelligent detection system and method based on image processing algorithm | |
CN113705564B (en) | Pointer type instrument identification reading method | |
CN110579184A (en) | Product appearance online detection device and use method thereof | |
CN114372984A (en) | Super-resolution component angle identification device and method | |
CN114913086B (en) | Face image quality enhancement method based on generation countermeasure network | |
CN116091506A (en) | Machine vision defect quality inspection method based on YOLOV5 | |
CN114332069B (en) | Connector detection method and device based on machine vision | |
CN115731195A (en) | Defect detection method, device, equipment and system |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
RJ01 | Rejection of invention patent application after publication |
Application publication date: 20220419 |
|
RJ01 | Rejection of invention patent application after publication |