CN102901735A - System for carrying out automatic detections upon workpiece defect, cracking, and deformation by using computer - Google Patents
System for carrying out automatic detections upon workpiece defect, cracking, and deformation by using computer Download PDFInfo
- Publication number
- CN102901735A CN102901735A CN2012103133311A CN201210313331A CN102901735A CN 102901735 A CN102901735 A CN 102901735A CN 2012103133311 A CN2012103133311 A CN 2012103133311A CN 201210313331 A CN201210313331 A CN 201210313331A CN 102901735 A CN102901735 A CN 102901735A
- Authority
- CN
- China
- Prior art keywords
- workpiece
- result
- module
- fission
- damaged
- 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.)
- Granted
Links
Images
Landscapes
- Investigating Materials By The Use Of Optical Means Adapted For Particular Applications (AREA)
- Image Processing (AREA)
Abstract
The invention discloses a system for carrying out automatic detections upon workpiece defect, cracking, and deformation by using a computer. With a computer image identification technology, three flaws of a workpiece, such as defect, cracking and deformation, can be automatically identified. First, a sample is sampled; qualified product parameter thresholds are obtained through training results of previous qualified products, or qualified product parameters are obtained through spot training; and qualification conditions of subsequent inputted workpieces are determined with the thresholds or the parameters as standards. The invention provides different algorithms aiming at different flaws. With the consideration of the differences in workpiece photographing situations, parameter adaptive adjustments are added into the algorithms, such that different photographing light intensities can be adapted to. Detection rates and missing rates of the three algorithms are respectively higher than 95% and lower than 5%, such that the effect is better than all previous algorithms.
Description
Technical field
The present invention relates to field of computer technology, utilize specifically that computing machine is damaged to workpiece, fission, the system that automatically detects of distortion.
Background technology
At present, the robotization that the processing of domestic part can the basic guarantee manufacturing procedure, but for the part defect that occurs in the process, namely substandard product still depends on artificial screening in a large number.The disadvantage of this mode mainly contains:
Need very many manpowers.Along with the progressively rising of domestic human cost, the cost of part also can increase thereupon.
The subjectivity of artificial screening is strong.Has higher loss.
Lack the defective data support, the reason that analyzing defect produces is inconvenient to recall in part processing enterprise
Therefore, utilizing image recognition that part is automatically detected is the trend that develops at present.
Summary of the invention
The present invention is directed to above shortcomings in the prior art, provide a kind of computing machine that utilizes to damaged, fission, the system that automatically detects of distortion of workpiece, utilize computer vision technique, automatically detect damaged, distortion and the algorithm of fission and the automatic checkout system framework of realizing based on this algorithm of workpiece.When guaranteeing to detect as much as possible the flaw article, avoid the false retrieval certified products.
For achieving the above object, the technical solution adopted in the present invention is as follows:
A kind ofly utilize that computing machine is damaged to workpiece, fission, the system that automatically detects of distortion, comprise concentrated collection terminal, certified products parameter training module, the conventional data end for process, exclusive data end for process and final as a result integrate module, wherein:
-concentrate collection terminal, duty is to the workpiece data sampling, and carries out delivering to system behind the automatic classification and process.Because the position of workpiece, defect is different, must accomplish defective is sampled into picture at collection terminal.
-conventional data end for process is used for not specific component and not detection and the processing of ad-hoc location;
This part is mainly for detection and the processing of the damaged of integral body and fission.
-exclusive data control end is used for detecting and processing for the defective that specific component occurs;
This part detects and processes mainly for the defective in workpiece deformation.
-described conventional data end for process and exclusive data end for process are given as a result integrate module with the result, the error message of being responsible for processing and exporting workpiece by integrate module as a result.
Described certified products parameter training module is processed by certified products being carried out image, extract characteristic parameter, utilize law of great numbers, judge that it meets normal distribution, variance by 3 times of mean value plus-minuss obtains between the certified products parameter region, and all not can be considered unacceptable product in this parameter distribution.
Described conventional data control end by to the not specific component of input and not the image of ad-hoc location process, judge its whether unacceptable product.
Described conventional data end for process comprises with lower device: damaged judge module is used for judging whether this workpiece is damaged; The fission judge module is used for judging whether this workpiece fissions.
Damaged judge module in the conventional data end for process, at first according to the automatic computed segmentation parameter of the gray-scale map of photo current, the size of calculating largest connected zone by binaryzation obtains the area of workpiece, and compare with the parameter that obtains by certified products training, if less than judging that then workpiece is damaged.Then the result is sent to as a result integrate module.
Fission judge module in the conventional data end for process, mainly in the fission of surface of the work, namely the silver layer of surface of the work cracking exposes copper layer etc.The at first same saturation degree space according to picture of this module, obtain subregion easy to crack, because cracking part is different from the saturation degree space for the part of cracking, obvious black and white difference will can be found behind the image binaryzation, calculate the area in largest connected zone, compare with the parameter that obtains by the certified products training, if greater than this value, then think and fission.The result is directly sent to as a result integrate module.
Described exclusive data control end is by processing the image of specific component of input, judges its whether unacceptable product.
Described exclusive data end for process comprises the distortion judge module: be used for judging whether angular distortion of workpiece.
Distortion judge module in the exclusive data end for process mainly for the planarization in workpiece is judged namely whether have angular distortion.Because the image that this module needs generally is the side image of workpiece, so need to work in the exclusive data end for process.This module is at first according to input picture, image is divided into a plurality of parts, after utilizing the skeleton thinning algorithm to carry out refinement to each part, remove noise, utilize subsequently the hough conversion to obtain near linear for the skeleton of refinement, the slope of different piece straight line is carried out differential comparison, the result of the result who obtains and certified products training module relatively, if the absolute value of each differential seat angle greater than the parameter value of certified products, then is judged as angular distortion.Then the result is delivered to as a result integrate module.
When the present invention works, took the situation situation change was arranged if last time take situation and this time, and then first with concentrating collection terminal that the photo of certified products is issued certified products parameter training module, allowed it carry out parameter and automatically train.Complete or when not needing to carry out parameter training at parameter training, the native system life's work.First by concentrating collection terminal that workpiece is taken, and with the photo automatic classification, the whole workpiece of judging of needs is delivered to the conventional data end for process, and what will need to take privileged sites (being that angular distortion need to be taken workpiece knuckle part) delivers to the exclusive data end for process.The conventional data end for process by the parameter that built-in algorithm and before training obtain, is carried out damaged judgement and fission judgement to the workpiece photo respectively, then mistake is delivered to as a result integrate module if find mistake.The exclusive data end for process by the parameter that built-in algorithm and before training obtain, is carried out respectively the judgement of angular distortion to the workpiece photo, then mistake is delivered to as a result integrate module if find mistake.Final as a result integrate module is exported the defective type of workpiece.
Compared with prior art, the use of the invention the parameter adaptive design, utilize computer vision technique really to accomplish the function of automatic identification workpiece, defect, remedied the existing deficiency that workpiece, defect detects of manually carrying out.By great amount of samples is detected, utilization result of the present invention is quite outstanding, and every type defective all reaches verification and measurement ratio and the 5% following fallout ratio more than 95%.
Description of drawings
Below by accompanying drawing technical solution of the present invention is done a detailed description, by accompanying drawing technical solution of the present invention is described for clearer, following accompanying drawing has all adopted the background color with gray scale:
Fig. 1 is system framework figure of the present invention;
Fig. 2～Fig. 4 is that corresponding respectively to of collecting of the concentrated collection terminal of the embodiment of the invention is damaged, defect ware sample and the certified products sample graph of fission and angular distortion;
Fig. 5 is the gray-scale map of damaged defect ware among Fig. 2;
Fig. 6 is the binary map of Fig. 5;
Fig. 7 is the fission zone of Fig. 3 of extraction;
Fig. 8 is the binary map of the fission defect ware form one of Fig. 7;
Fig. 9 is the binary map of the fission defect ware form two of Fig. 7;
Figure 10 is the binary map of fission certified products in the embodiment of the invention;
Figure 11 is through the fission defect ware binary map after the scan-line algorithm extraction in the embodiment of the invention;
Figure 12 is the binary map of the distortion defect ware of Fig. 4;
Figure 13 is the string diagram that obtains behind Figure 12 process skeletal extraction algorithm;
Figure 14 is divided near linear figure after three parts are carried out respectively the hough conversion with Figure 13.
Embodiment
Below in conjunction with accompanying drawing embodiments of the invention are elaborated, present embodiment is implemented under take the invention technical scheme as prerequisite, provided detailed embodiment and concrete operating process, but protection scope of the present invention is not limited to following embodiment.
The task of present embodiment is workpiece to be carried out the detection of three types.
As shown in Figure 1, the present invention includes 5 modules: concentrate collection terminal, certified products parameter training module, conventional data end for process, exclusive data end for process and integrate module as a result.
By concentrating collection terminal, obtain three samples pictures, respectively as shown in Figure 2 corresponding to damaged defect ware sample (the right figure among Fig. 2) and certified products sample (the left figure among Fig. 2), as shown in Figure 3 defect ware sample (the right figure among Fig. 3) and certified products sample (the left figure among Fig. 3) and the defect ware sample corresponding to angular distortion as shown in Figure 4 (the right figure among Fig. 4) and certified products sample (the left figure among Fig. 4) corresponding to fission.
As shown in Figure 2, left figure is certified products, and right figure has defect ware.It is damaged that right figure can see significantly that the lower left corner has.The below is described with the flow process of legend for damaged product.
As shown in Figure 5, this figure is the gray-scale map of the defect ware among Fig. 2, by extracting the mean value of gray-scale map, multiply by the fixed constant after a certain training, through after the binaryzation, obtains binary map as shown in Figure 6.The reason that extraction mean value multiply by fixed constant is, prevents that the gray-scale value deviation that causes is owing to each gray-scale value even variation, admittedly use the method can accomplish the adjustment of binaryzation threshold self-adaptation owing to take the light difference.Calculate largest connected zone according to the binary map that obtains as shown in Figure 6 this moment, the number of largest connected white pixel, and the parameter that obtains with before certified products training compares, and then can be judged to be damaged product.
As shown in Figure 7, this figure is the fission zone that extracts according to workpiece size, in fact, if the fission zone is positioned at whole workpiece, also can not extract the zone, in this example, because fission only is present on the silver layer of workpiece, strengthens specific aim so extract the zone.Utilize equally average to multiply by the binary map that obtains after the threshold of fixed constant as binaryzation as shown in Figure 8.Here we need to extract real fission zone, and reason be the to fission lacquer on upper strata normally falls to exposing the color of bottom, and bottom is copper in this example, and what expose is copper equally, so thereby need to extract the silver layer position removes incoherent copper face.What take here is scan-line algorithm, both begun from the bottom up with one with sweep trace scanning, when below this scanning is satisfied through line during one of two conditions, we think that this line is the cut-off rule of silver-colored face and copper face: 1. the pixel of the unnecessary white of pixel of black; 2. the piece number of black region is greater than a certain predetermined value.If do not want to connect if this algorithm is considered cracking part and copper face, as shown in Figure 8, then scheme 1 satisfies.If the cracking part links to each other with copper face, as shown in Figure 9, then scheme 2 satisfies.The correctness of this algorithm is based on the smooth smooth and desultory phenomenon of copper of cracking part of the copper in the complete copper face.From test result, this algorithm is effective.As shown in figure 10, the silver-colored face portion of certified products is smooth smooth.As shown in figure 11, calculate the area of the middle white in the cross section that extracts through scan-line algorithm, if fission greater than training the value that obtains to think by certified products.
As shown in figure 12, this figure is through the binary map after the self adaptive adjustment of threshold.Figure 13 passes through the result who obtains behind skeletal extraction algorithm for this binary map.Shown in Figure 14, the lines among Figure 13 are divided into three parts, carry out respectively the hough(Hough) and conversion, the straight line of grey color part for extracting.The angle of the straight line that obtains is subtracted each other, and compare with the parameter of certified products, can find the third part among Figure 14, obviously do not become 90 degree with second portion among Figure 14, then think angular distortion.
At last the result is all delivered to as a result integrate module, the type of as a result integrate module output error.
More than specific embodiments of the invention are described.It will be appreciated that the present invention is not limited to above-mentioned particular implementation, those skilled in the art can make various distortion or modification within the scope of the claims, and this does not affect flesh and blood of the present invention.
Claims (5)
- One kind utilize that computing machine is damaged to workpiece, fission, the system that automatically detects of distortion, it is characterized in that, comprise concentrated collection terminal, certified products parameter training module, conventional data control end, exclusive data control end and final as a result integrate module, wherein:-concentrate collection terminal, be responsible for the workpiece data sampling, and carry out delivering to system behind the automatic classification and process;-certified products parameter training module is used for the parameters of certified products is trained automatically;-conventional data control end, be used for to the not specific component of input and not the image of ad-hoc location process, judge its whether unacceptable product;-exclusive data control end is used for the image of the specific component of input is processed, and judges its whether unacceptable product;-integrate module as a result, described conventional data end for process and exclusive data end for process are given as a result integrate module with the result, the error message of being responsible for processing and exporting workpiece by integrate module as a result.
- 2. according to claim 1ly utilize that computing machine is damaged to workpiece, fission, the system that automatically detects of distortion, it is characterized in that, described certified products parameter training module is to process by certified products being carried out image, extract characteristic parameter, utilize law of great numbers, judge that it meets normal distribution, the variance by 3 times of mean value plus-minuss obtains between the certified products parameter region, and all not can be considered unacceptable product in this parameter distribution.
- 3. according to claim 1ly utilize that computing machine is damaged to workpiece, fission, the system that automatically detects of distortion, it is characterized in that described conventional data control end comprises with lower device:Damaged judge module is used for judging whether this workpiece is damaged;Described damaged judge module is at first according to the automatic computed segmentation parameter of the gray-scale map of photo current, the size of calculating largest connected zone by binaryzation obtains the area of workpiece, and compare with the parameter that obtains by certified products training, if less than judging that then workpiece is damaged, the result is sent to as a result integrate module;The fission judge module is used for judging whether this workpiece fissions;Described fission judge module is mainly in the fission of surface of the work, at first according to the saturation degree space of picture, obtain subregion easy to crack, because the cracking part is different from the saturation degree space of the part that do not ftracture, with finding obvious black and white difference behind the image binaryzation, calculate the area in largest connected zone, compare with the parameter that obtains by the certified products training, if greater than this value, then think and fission, the result is directly sent to as a result integrate module.
- 4. according to claim 1ly utilize that computing machine is damaged to workpiece, fission, the system that automatically detects of distortion, it is characterized in that described exclusive data control end comprises the distortion judge module, be used for judging whether angular distortion of workpiece;Described distortion judge module is at first according to input picture, image is divided into a plurality of parts, after utilizing the skeleton thinning algorithm to carry out refinement to each part, remove noise, utilize subsequently the hough conversion to obtain near linear for the skeleton of refinement, the slope of different piece straight line is carried out differential comparison, the result of the result who obtains and certified products training module relatively, if the absolute value of each differential seat angle greater than the parameter value of certified products, then is judged as angular distortion, the result is delivered to as a result integrate module.
- 5. according to claim 1ly utilize that computing machine is damaged to workpiece, fission, the system that automatically detects of distortion, it is characterized in that, described as a result integrate module is for the problem of all existence of output workpiece and with damaged judge module, and the result of fission judge module and distortion judge module integrates simultaneously output.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201210313331.1A CN102901735B (en) | 2012-08-29 | 2012-08-29 | System for carrying out automatic detections upon workpiece defect, cracking, and deformation by using computer |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201210313331.1A CN102901735B (en) | 2012-08-29 | 2012-08-29 | System for carrying out automatic detections upon workpiece defect, cracking, and deformation by using computer |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN102901735A true CN102901735A (en) | 2013-01-30 |
| CN102901735B CN102901735B (en) | 2015-02-25 |
Family
ID=47574088
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201210313331.1A Active CN102901735B (en) | 2012-08-29 | 2012-08-29 | System for carrying out automatic detections upon workpiece defect, cracking, and deformation by using computer |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN102901735B (en) |
Cited By (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104048968A (en) * | 2014-06-04 | 2014-09-17 | 上海美迪索科电子科技有限公司 | Industrial processing part automatic defect identification system |
| CN104197836A (en) * | 2014-09-13 | 2014-12-10 | 江南大学 | Vehicle lock assembly size detection method based on machine vision |
| CN108693196A (en) * | 2017-04-11 | 2018-10-23 | 欧姆龙株式会社 | Sheet material check device |
| CN109255787A (en) * | 2018-10-15 | 2019-01-22 | 杭州慧知连科技有限公司 | Silk ingot scratch detection system and method based on deep learning and image processing techniques |
| CN109463894A (en) * | 2018-12-27 | 2019-03-15 | 蒋梦兰 | Configure the full water-proof type toothbrush of half-moon-shaped brush head |
| CN111667094A (en) * | 2020-04-22 | 2020-09-15 | 深圳市吉迩科技有限公司 | Automatic detection method, system and device |
| CN112720500A (en) * | 2020-12-30 | 2021-04-30 | 深兰人工智能芯片研究院（江苏）有限公司 | Control method and device for manipulator, pickup device and storage medium |
Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JPS56161651A (en) * | 1980-05-15 | 1981-12-12 | Mitsubishi Electric Corp | Malfunction analysing method of semiconductor element |
| JP2003329596A (en) * | 2002-05-10 | 2003-11-19 | Mitsubishi Rayon Co Ltd | Apparatus and method for inspecting defect |
| CN101221135A (en) * | 2008-01-17 | 2008-07-16 | 中国电子科技集团公司第四十五研究所 | Printed circuit board image skeletonization method based on FPGA |
| CN201229307Y (en) * | 2008-07-18 | 2009-04-29 | 深圳市联君科技有限公司 | Novel cigarette package intelligent detection device |
| CN101799431A (en) * | 2010-04-08 | 2010-08-11 | 湖南大学 | Machine visual on-line detection method and equipment for tank cover quality in high-speed production line |
| CN102095733A (en) * | 2011-03-02 | 2011-06-15 | 上海大学 | Machine vision-based intelligent detection method for surface defect of bottle cap |
| CN102253050A (en) * | 2011-03-14 | 2011-11-23 | 广州市盛通建设工程质量检测有限公司 | Automatic detection method and device for magnetic tile surface defect based on machine vision |
| CN102621156A (en) * | 2012-03-07 | 2012-08-01 | 武汉理工大学 | Image-processing-based automatic micro part sorting system |
-
2012
- 2012-08-29 CN CN201210313331.1A patent/CN102901735B/en active Active
Patent Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JPS56161651A (en) * | 1980-05-15 | 1981-12-12 | Mitsubishi Electric Corp | Malfunction analysing method of semiconductor element |
| JP2003329596A (en) * | 2002-05-10 | 2003-11-19 | Mitsubishi Rayon Co Ltd | Apparatus and method for inspecting defect |
| CN101221135A (en) * | 2008-01-17 | 2008-07-16 | 中国电子科技集团公司第四十五研究所 | Printed circuit board image skeletonization method based on FPGA |
| CN201229307Y (en) * | 2008-07-18 | 2009-04-29 | 深圳市联君科技有限公司 | Novel cigarette package intelligent detection device |
| CN101799431A (en) * | 2010-04-08 | 2010-08-11 | 湖南大学 | Machine visual on-line detection method and equipment for tank cover quality in high-speed production line |
| CN102095733A (en) * | 2011-03-02 | 2011-06-15 | 上海大学 | Machine vision-based intelligent detection method for surface defect of bottle cap |
| CN102253050A (en) * | 2011-03-14 | 2011-11-23 | 广州市盛通建设工程质量检测有限公司 | Automatic detection method and device for magnetic tile surface defect based on machine vision |
| CN102621156A (en) * | 2012-03-07 | 2012-08-01 | 武汉理工大学 | Image-processing-based automatic micro part sorting system |
Cited By (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104048968A (en) * | 2014-06-04 | 2014-09-17 | 上海美迪索科电子科技有限公司 | Industrial processing part automatic defect identification system |
| CN104197836A (en) * | 2014-09-13 | 2014-12-10 | 江南大学 | Vehicle lock assembly size detection method based on machine vision |
| CN108693196A (en) * | 2017-04-11 | 2018-10-23 | 欧姆龙株式会社 | Sheet material check device |
| CN109255787A (en) * | 2018-10-15 | 2019-01-22 | 杭州慧知连科技有限公司 | Silk ingot scratch detection system and method based on deep learning and image processing techniques |
| CN109463894A (en) * | 2018-12-27 | 2019-03-15 | 蒋梦兰 | Configure the full water-proof type toothbrush of half-moon-shaped brush head |
| CN111667094A (en) * | 2020-04-22 | 2020-09-15 | 深圳市吉迩科技有限公司 | Automatic detection method, system and device |
| CN112720500A (en) * | 2020-12-30 | 2021-04-30 | 深兰人工智能芯片研究院（江苏）有限公司 | Control method and device for manipulator, pickup device and storage medium |
Also Published As
| Publication number | Publication date |
|---|---|
| CN102901735B (en) | 2015-02-25 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN102901735B (en) | System for carrying out automatic detections upon workpiece defect, cracking, and deformation by using computer | |
| CN109816644B (en) | Bearing defect automatic detection system based on multi-angle light source image | |
| CN108230315B (en) | Mask belt loss detection method based on machine vision | |
| CN109859181A (en) | A kind of PCB welding point defect detection method | |
| Yazdi et al. | Feature extraction algorithm for fill level and cap inspection in bottling machine | |
| CN105046700A (en) | Brightness correction and color classification-based fruit surface defect detection method and system | |
| CN108918093B (en) | Optical filter mirror surface defect detection method and device and terminal equipment | |
| CN107490582A (en) | A kind of streamline Work Piece Verification System Based | |
| CN109461156B (en) | Threaded sealing plug assembly detection method based on vision | |
| CN107490583A (en) | A kind of intermediate plate defect inspection method based on machine vision | |
| CN111161237A (en) | Fruit and vegetable surface quality detection method, storage medium and sorting device thereof | |
| CN111122590A (en) | Ceramic surface defect detection device and detection method | |
| WO2023168984A1 (en) | Area-array camera-based quality inspection method and system for cathode copper | |
| Pithadiya et al. | Comparison of optimal edge detection algorithms for liquid level inspection in bottles | |
| CN114820626B (en) | Intelligent detection method for automobile front face part configuration | |
| CN113894055A (en) | Hardware surface defect detection and classification system and method based on machine vision | |
| KR100687811B1 (en) | Method and apparatus for detecting defects of the welded part | |
| CN115953373A (en) | Glass defect detection method and device, electronic equipment and storage medium | |
| CN106353340A (en) | Surface defect detection method for rod-like high-reflectance part | |
| CN113019973A (en) | Online visual inspection method for manufacturing defects of ring-pull cans | |
| CN115866502A (en) | Microphone part surface defect online detection process | |
| WO2022222467A1 (en) | Open circular ring workpiece appearance defect detection method and system, and computer storage medium | |
| CN115170501A (en) | Defect detection method, system, electronic device and storage medium | |
| CN109682821B (en) | Citrus surface defect detection method based on multi-scale Gaussian function | |
| CN111724375B (en) | Screen detection method and system |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C14 | Grant of patent or utility model | ||
| GR01 | Patent grant |