CN109829906A - It is a kind of based on the workpiece, defect of the field of direction and textural characteristics detection and classification method - Google Patents
It is a kind of based on the workpiece, defect of the field of direction and textural characteristics detection and classification method Download PDFInfo
- Publication number
- CN109829906A CN109829906A CN201910095715.2A CN201910095715A CN109829906A CN 109829906 A CN109829906 A CN 109829906A CN 201910095715 A CN201910095715 A CN 201910095715A CN 109829906 A CN109829906 A CN 109829906A
- Authority
- CN
- China
- Prior art keywords
- image
- workpiece
- defect
- field
- sample
- 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
Abstract
The invention discloses a kind of based on the detection of the workpiece, defect of the field of direction and textural characteristics and classification method, including establishing SVM detection model first with the workpiece sample data of acquisition, the workpiece for measurement surface image of acquisition is carried out to pre-process the automatic ROI region that must obtain workpiece, ROI region is divided into the sample to be tested image of w × w size, and label is added in sample to be tested image coordinate according to position coordinates, finally classified with SVM detection model to sample to be tested image, classification results and defect sample are obtained, defective locations are further indexed by label.The present invention calculates contrast, the feature vector of energy, entropy in the field of direction and textural characteristics of image, scale conversion is carried out by normalization, to establish SVM model, the model of calculating has not only reacted the feature of different defects, influence of the ambient lighting to system can be reduced simultaneously, increases the stability and anti-interference ability of system.
Description
Technical field
The present invention relates to workpiece, defect detection technique field more particularly to a kind of workpiece based on the field of direction and textural characteristics
Defects detection and classification method.
Background technique
Workpiece, defect detection is always the important research topic of manufacture field, can bring direct economic benefit, greatly
Liberate the productive forces.In recent years, machine vision technique is quickly grown, but surface defects detection is always manufacturing difficult point.
It is primarily due to the feature of image difference that workpiece surface reflective character is different, and different polishing modes acquires, needs to set
Different defects detection algorithms is counted, therefore surface defects detection is always industry pain spot.For high reflection characteristic workpiece surface
Defects detection is particularly difficult, is primarily due to the reflective acquisition image that will cause of mirror surface spike caused by high reflection characteristic and generates
Dazzle causes difficulty to image algorithm.
The principle that defects detection is realized using streak reflex is based on phase deviation technology, using the light of Sine Modulated
Source, the image of multi collect out of phase, solution phase realize three-dimensional surface modeling, to judge defect, such method is suitable for
Ideal mirror or optical element, in high reflection workpiece, defect detection occasion by dust, the interference such as reflection light intensity are difficult to reach
Classify to perfect precision and real-time detection
Summary of the invention
It detects and divides based on the workpiece, defect of the field of direction and textural characteristics in view of this, the object of the present invention is to provide a kind of
Class method can be taken using stripe pattern into high reflection face catoptric imaging using the display screen of display stripe pattern as area source
Band is detected the characteristic of planar defect information to realize Image Acquisition.
The present invention solves above-mentioned technical problem by following technological means:
It is a kind of based on the workpiece, defect of the field of direction and textural characteristics detection and classification method, comprising:
S1: building image capturing system, acquires workpiece for measurement surface image, is pre-processed the surface image to obtain
Take the ROI region of workpiece for measurement；
S2: the ROI region for obtaining workpiece for measurement corresponds to the characteristic parameter of defect sample image, establishes and is based on field of direction vector
With the SVM detection model of texture feature vector；
The ROI region of sample to be tested: being divided into the sample to be tested image of w × w size by S3, and according to position coordinates by institute
It states sample to be tested image coordinate and label is added, wherein w is the width dimensions of sample to be tested image；
S4: classified with the SVM detection model to the sample to be tested image, obtain classification results and defect sample
This；
S5: the coordinate information in the label of the classification results and the defect sample is obtained, the classification results are stored
With the coordinate information of the defect sample；
S6: judging whether the workpiece for measurement detects completion, exports the classification results if detecting and completing and controls
Executing agency executes corresponding actions；If detection does not complete and thens follow the steps S2-S6, until detection is completed.
Further, the characteristic parameter for obtaining workpiece and corresponding to defect image, is established special based on field of direction vector sum texture
The method of SVM detection model for levying vector includes:
Prepare workpiece and correspond to the data set of defect sample image, and the data set is divided into training set and test set；
The field of direction vector characteristics parameter and texture of defect sample image are corresponded to according to the workpiece in the training set
Feature vector characteristic parameter establishes initial SVM detection model；
Defect sample image analysis selection kernel function is corresponded to according to the workpiece in the training set；
Cross validation is carried out, the punishment parameter C of the initial SVM detection model and the parameter g of kernel function are selected；
Punishment parameter C and kernel function described in the initial SVM detection model and adjusting and optimizing are tested using the test set
Parameter g, establish the SVM detection model.
Further, described the step of pre-processing the automatic ROI region that obtain workpiece carried out to the surface image to include:
Gray processing is carried out to the surface image, obtains gray level image；
The histogram of the gray level image is calculated, automatic global threshold segmentation is carried out based on maximum entropy, by the grayscale image
As being divided into piecemeal sample image, the ROI region of the piecemeal sample image is obtained automatically by minimax area.
The ROI region using Gaussian transformation and is transformed into airspace in frequency domain, it is poor to carry out with the piecemeal sample image
Partite transport is calculated, and is realized that the defect characteristic of the piecemeal sample image enhances by histogram equalization, is obtained enhancing image.
Further, the method for building up of the field of direction feature vector includes:
Estimate the field of direction of the enhancing image, using lowest mean square direction estimation algorithm to extract the enhancing image
Field of direction feature vector；
The textural characteristics for calculating the enhancing image, establish field of direction feature vector set.
Further, the method for building up of the texture feature vector includes:
Calculate the gray level co-occurrence matrixes of the piecemeal sample image；
Use contrast, energy and the entropy of gray level co-occurrence matrixes as texture feature vector.
Further, after obtaining the field of direction feature vector and the texture feature vector, to the field of direction feature to
Amount is normalized with the texture feature vector.
Further, the classification of the defect sample image is divided into sand holes, pit, abrasive band trace, dirty and standard zero defect.
Beneficial effects of the present invention:
The present invention calculates contrast, the feature vector of energy, entropy in the field of direction and textural characteristics of image, passes through normalization
Scale conversion is carried out, to establish SVM model, chooses the feature for not only having reacted different defects, while ambient lighting can be reduced
Influence to system increases the stability and anti-interference ability of system.
When handling workpiece original image, since original image is excessive, it is unfavorable for algorithm process.The present invention will be former
The regular division of beginning image is blocking, and is numbered, and establishes numeral index.The position of image can be obtained by numbering.?
After obtaining classification results, the position of available defect is numbered using it, is avoided and is gone positioning defect using a large amount of algorithm, it is fixed
Position precision is related with picture size is divided, and meets industrial detection demand, but greatly improves the real-time of system.
Detailed description of the invention
Fig. 1 is a kind of flow chart based on the workpiece, defect of the field of direction and textural characteristics detection and classification method of the present invention；
Fig. 2 is the flow chart that the present invention establishes SVM detection model.
Specific embodiment
Below with reference to the drawings and specific embodiments, the present invention is described in detail:
As shown in Figure 1, of the invention is a kind of based on the detection of the workpiece, defect of the field of direction and textural characteristics and classification method, packet
It includes:
S1: building image capturing system, acquires workpiece for measurement surface image, is pre-processed the surface image to obtain
Take the ROI region of workpiece for measurement；
S2: the ROI region for obtaining workpiece for measurement corresponds to the characteristic parameter of defect sample image, establishes and is based on field of direction vector
With the SVM detection model of texture feature vector；
The ROI region of sample to be tested: being divided into the sample to be tested image of w × w size by S3, and according to position coordinates by institute
It states sample to be tested image coordinate and label is added, wherein w is the width dimensions of sample to be tested image；
S4: classified with the SVM detection model to the sample to be tested image, obtain classification results and defect sample
This；
S5: the coordinate information in the label of the classification results and the defect sample is obtained, the classification results are stored
With the coordinate information of the defect sample；
S6: judging whether the workpiece for measurement detects completion, exports the classification results if detecting and completing and controls
Executing agency executes corresponding actions；If detection does not complete and thens follow the steps S2-S6, until detection is completed.
Specifically, as shown in Fig. 2, the workpiece that obtains correspond to the characteristic parameter of defect image, establish be based on the field of direction to
Amount and the method for the SVM detection model of texture feature vector include:
Prepare workpiece and correspond to the data set of defect sample image, and the data set is divided into training set and test set；
The field of direction vector characteristics parameter and texture of defect sample image are corresponded to according to the workpiece in the training set
Feature vector characteristic parameter establishes initial SVM detection model；
Defect sample image analysis selection kernel function is corresponded to according to the workpiece in the training set；
Cross validation is carried out, the punishment parameter C of the initial SVM detection model and the parameter g of kernel function are selected；
Punishment parameter C and kernel function described in the initial SVM detection model and adjusting and optimizing are tested using the test set
Parameter g, establish the SVM detection model.
Optionally, described the step of pre-processing the automatic ROI region that obtain workpiece carried out to the surface image to include:
Gray processing is carried out to the surface image, obtains gray level image；
The histogram of the gray level image is calculated, automatic global threshold segmentation is carried out based on maximum entropy, by the grayscale image
As being divided into piecemeal sample image, the ROI region of the piecemeal sample image is obtained automatically by minimax area.
Specifically, the RGB color image of acquisition is converted into gray level image g (x, y)=0.299*R+0.587*G+0.114B,
Workpiece for measurement will become clear under light illumination than background, therefore count to the gray scale of image, the formula of statistics are as follows:
In formula, n is the sum of pixel in image, nxIt is the pixel quantity that gray level is x, L is the possible gray scale of image
Grade sum, 0≤L≤255.Calculate the comentropy of gray level image:
Hx=-p (x) log p (x) dx
For by altimetric image, workpiece for measurement is in target area, monitor station is then background area, the maximum gradation value of image
Grade is L.It then finds out target area and obtains entropy are as follows:
Background area obtains entropy
Solve each gray level t, w (t)=H0+H1, maximum entropy vector t* satisfaction:
W(t*)=max (W (t))
After dividing the image into different regions, feature detection is carried out using area features value, automatic can must obtain work
Part obtains ROI region.
The ROI region using Gaussian transformation and is transformed into airspace in frequency domain, it is poor to carry out with the piecemeal sample image
Partite transport is calculated, and by realizing that the background luminance of image is estimated, and can be weakened striped background after calculus of differences and be caused to defect information
Interference, by histogram equalization realize the piecemeal sample image defect characteristic enhance, obtain enhancing image.
Further, the method for building up of the field of direction feature vector includes:
Estimate the field of direction of the enhancing image, using lowest mean square direction estimation algorithm to extract the enhancing image
Field of direction feature vector；
The textural characteristics for calculating the enhancing image, establish field of direction feature vector set.
Specifically, the field of direction describes a two-dimensional surface field in streakline direction and corresponding position in image, image direction
Field size is the value of gradient fields Orthogonal Decomposition parameter θ, and right hand theorem is pressed perpendicular to gradient direction in direction.Assuming that into after crossing step 2
Image is G, carries out being blocked into w × w block to image by equal size.The gradient of image is calculated according to pixelWithFoundation point direction calculating direction obtained in the previous step.Calculation formula is as follows:
Wherein θ (i, j) is the least mean-square estimate of local Block direction,It is image f (x, y) in step 2 in picture
To the local derviation of x at vegetarian refreshments (x, y).For image f (x, y) in step 2 to the local derviation of y at pixel (x, y).
Since the direction of current block is more slow than meeting with overall region direction, can be made using a low-pass filter
There is certain adjustment in incorrect direction.Directional diagram is transformed into a continuous vector field.Conversion formula is as follows:
φx(i, j)=cos (2 θ (i, j)), φy(i, j)=sin (2 θ (i, j))
Wherein φxAnd φyIt is the x and y-component of vector field respectively.Low-pass filtering calculation formula is as follows:
It is W that wherein W, which is window size,φ×WφTwo-dimensional low-pass filter, orientation field computation result is obtained after calculating:
After the field of direction vector calculated, as a feature vectors sample.
Further, the method for building up of the texture feature vector includes:
Calculate piecemeal sample gray level co-occurrence matrixes as additional eigenvectors, gray level co-occurrence matrixes include contrast,
Energy, entropy, inverse variance, correlative character, have used first three feature as useful feature in the present embodiment.Calculation formula point
It is not as follows:
The value of contrast metric metric matrix is the number how being distributed with localized variation in image, has reacted the clear of image
The rill depth of clear degree and texture, calculation formula are as follows:
Energy conversion reflects image grayscale and is evenly distributed degree and texture fineness degree.If the element value of gray level co-occurrence matrixes
Close, then energy is smaller, indicates that texture is careful；If some of them value is big, and other values are small, then energy value is larger.Energy value is big
Show a kind of texture pattern of more uniform and regular variation.Its calculation formula is as follows:
Image includes the randomness metrics of information content.When all values are equal in co-occurrence matrix or pixel value is shown most
When big randomness, entropy is maximum；Therefore entropy shows the complexity of image grayscale distribution, and entropy is bigger, and image is more complicated.
Its calculation formula is as follows:
Further, after obtaining the field of direction feature vector and the texture feature vector, due to different feature vectors
Physical significance it is different, therefore its value range is also different, therefore vector is normalized, and makes feature vector
Value is all fallen between zero and one.By normalized, the anti-interference ability of overall model can be enhanced.
Specifically, the classification of the defect sample image is divided into sand holes, pit, abrasive band trace, dirty and standard zero defect.
The above examples are only used to illustrate the technical scheme of the present invention and are not limiting, although referring to preferred embodiment to this hair
It is bright to be described in detail, those skilled in the art should understand that, it can modify to technical solution of the present invention
Or equivalent replacement should all cover without departing from the objective and range of technical solution of the present invention in claim of the invention
In range.Technology not described in detail in the present invention, shape, construction portion are well-known technique.
Claims (7)
1. a kind of based on the detection of the workpiece, defect of the field of direction and textural characteristics and classification method characterized by comprising
S1: building image capturing system, acquires workpiece for measurement surface image, the surface image is pre-processed with obtain to
Survey the ROI region of workpiece；
S2: the ROI region for obtaining workpiece for measurement corresponds to the characteristic parameter of defect sample image, establishes and is based on field of direction vector sum line
Manage the SVM detection model of feature vector；
The ROI region of sample to be tested: being divided into the sample to be tested image of w × w size by S3, and according to position coordinates will it is described to
Label is added in this image coordinate of test sample, wherein w is the width dimensions of sample to be tested image；
S4: classified with the SVM detection model to the sample to be tested image, obtain classification results and defect sample；
S5: the coordinate information in the label of the classification results and the defect sample is obtained, the classification results and institute are stored
State the coordinate information of defect sample；
S6: judging whether the workpiece for measurement detects completion, exports the classification results if detecting and completing and controls execution
Mechanism executes corresponding actions；If detection does not complete and thens follow the steps S2-S6, until detection is completed.
2. it is according to claim 1 it is a kind of based on the workpiece, defect of the field of direction and textural characteristics detection and classification method,
It is characterized in that, the characteristic parameter for obtaining workpiece for measurement and corresponding to defect sample image is established and is based on field of direction vector sum texture
The method of the SVM detection model of feature vector includes:
Prepare workpiece and correspond to the data set of defect sample image, and the data set is divided into training set and test set；
The field of direction vector characteristics parameter and texture feature vector of defect sample image are corresponded to according to the workpiece in the training set
Characteristic parameter establishes initial SVM detection model；
Defect sample image analysis selection kernel function is corresponded to according to the workpiece in the training set；
Cross validation is carried out, the punishment parameter C of the initial SVM detection model and the parameter g of kernel function are selected；
The ginseng of punishment parameter C and kernel function described in the initial SVM detection model and adjusting and optimizing are tested using the test set
Number g, establishes the SVM detection model.
3. it is according to claim 2 it is a kind of based on the workpiece, defect of the field of direction and textural characteristics detection and classification method,
It is characterized in that, it is described the step of pre-processing the automatic ROI region that obtain workpiece carried out to the surface image to include:
Gray processing is carried out to the surface image, obtains gray level image；
The histogram of the gray level image is calculated, automatic global threshold segmentation is carried out based on maximum entropy, by the gray level image point
It is segmented into piecemeal sample image, obtains the ROI region of the piecemeal sample image automatically by minimax area；
The ROI region using Gaussian transformation and is transformed into airspace in frequency domain, carries out difference fortune with the piecemeal sample image
It calculates, realizes that the defect characteristic of the piecemeal sample image enhances by histogram equalization, obtain enhancing image.
4. it is according to claim 3 it is a kind of based on the workpiece, defect of the field of direction and textural characteristics detection and classification method,
It is characterized in that, the method for building up of the field of direction feature vector includes:
Estimate the field of direction of the enhancing image, using lowest mean square direction estimation algorithm to extract the direction of the enhancing image
Field feature vector；
The textural characteristics for calculating the enhancing image, establish field of direction feature vector set.
5. it is according to claim 4 it is a kind of based on the workpiece, defect of the field of direction and textural characteristics detection and classification method,
It is characterized in that, the method for building up of the texture feature vector includes:
Calculate the gray level co-occurrence matrixes of the piecemeal sample image；
Use contrast, energy and the entropy of gray level co-occurrence matrixes as texture feature vector.
6. it is according to claim 5 it is a kind of based on the workpiece, defect of the field of direction and textural characteristics detection and classification method,
It is characterized in that: after obtaining the field of direction feature vector and the texture feature vector, to the field of direction feature vector and institute
Texture feature vector is stated to be normalized.
7. it is according to claim 6 it is a kind of based on the workpiece, defect of the field of direction and textural characteristics detection and classification method,
Be characterized in that: the classification of the defect sample image is divided into sand holes, pit, abrasive band trace, dirty and standard zero defect.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910095715.2A CN109829906A (en) | 2019-01-31 | 2019-01-31 | It is a kind of based on the workpiece, defect of the field of direction and textural characteristics detection and classification method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910095715.2A CN109829906A (en) | 2019-01-31 | 2019-01-31 | It is a kind of based on the workpiece, defect of the field of direction and textural characteristics detection and classification method |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN109829906A true CN109829906A (en) | 2019-05-31 |
Family
ID=66862896
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910095715.2A Pending CN109829906A (en) | 2019-01-31 | 2019-01-31 | It is a kind of based on the workpiece, defect of the field of direction and textural characteristics detection and classification method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109829906A (en) |
Cited By (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110414595A (en) * | 2019-07-25 | 2019-11-05 | 广西科技大学 | The orientation estimate method of texture image with orientation consistency |
| CN111353992A (en) * | 2020-03-10 | 2020-06-30 | 塔里木大学 | Agricultural product defect detection method and system based on textural features |
| CN111539939A (en) * | 2020-04-27 | 2020-08-14 | 扬州哈工科创机器人研究院有限公司 | Defect detection method and device based on machine vision |
| CN111862080A (en) * | 2020-07-31 | 2020-10-30 | 易思维（杭州）科技有限公司 | Deep learning defect identification method based on multi-feature fusion |
| CN112200789A (en) * | 2020-10-16 | 2021-01-08 | 中国铁道科学研究院集团有限公司 | Image identification method and device, electronic equipment and storage medium |
| CN113689427A (en) * | 2021-10-25 | 2021-11-23 | 常州微亿智造科技有限公司 | Defect detection method for parameter optimization based on space particle automatic attraction algorithm |
| CN115187590A (en) * | 2022-09-08 | 2022-10-14 | 山东艾克赛尔机械制造有限公司 | Automobile part defect detection method based on machine vision |
| CN115619767A (en) * | 2022-11-09 | 2023-01-17 | 南京云创大数据科技股份有限公司 | Method and device for detecting surface defects of mirror-like workpiece based on multi-illumination condition |
| CN117237340A (en) * | 2023-11-10 | 2023-12-15 | 江西省中鼐科技服务有限公司 | Method and system for detecting appearance of mobile phone shell based on artificial intelligence |
| CN117314914A (en) * | 2023-11-29 | 2023-12-29 | 广州市市政工程试验检测有限公司 | Defect identification method for engineering nondestructive testing image and related equipment |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105261003A (en) * | 2015-09-10 | 2016-01-20 | 西安工程大学 | Defect point detection method on basis of self structure of fabric |
| CN105466959A (en) * | 2016-01-07 | 2016-04-06 | 安阳中科工程检测有限公司 | Method for positioning defects in large detected piece nondestructive detection |
| CN107123114A (en) * | 2017-04-21 | 2017-09-01 | 佛山市南海区广工大数控装备协同创新研究院 | A kind of cloth defect inspection method and device based on machine learning |
| CN107525813A (en) * | 2017-09-21 | 2017-12-29 | 常州有恒智能装备科技有限公司 | Steel plate defect on-line intelligence vision detection system |
| CN108765402A (en) * | 2018-05-30 | 2018-11-06 | 武汉理工大学 | Non-woven fabrics defects detection and sorting technique |
-
2019
- 2019-01-31 CN CN201910095715.2A patent/CN109829906A/en active Pending
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105261003A (en) * | 2015-09-10 | 2016-01-20 | 西安工程大学 | Defect point detection method on basis of self structure of fabric |
| CN105466959A (en) * | 2016-01-07 | 2016-04-06 | 安阳中科工程检测有限公司 | Method for positioning defects in large detected piece nondestructive detection |
| CN107123114A (en) * | 2017-04-21 | 2017-09-01 | 佛山市南海区广工大数控装备协同创新研究院 | A kind of cloth defect inspection method and device based on machine learning |
| CN107525813A (en) * | 2017-09-21 | 2017-12-29 | 常州有恒智能装备科技有限公司 | Steel plate defect on-line intelligence vision detection system |
| CN108765402A (en) * | 2018-05-30 | 2018-11-06 | 武汉理工大学 | Non-woven fabrics defects detection and sorting technique |
Non-Patent Citations (3)
| Title |
|---|
| HAILAN KUANG ET AL.: "Defect Detection of Bamboo Strips Based on LBP and GLCM Features by Using SVM Classifier", 《IEEE》 * |
| 李长云，王志兵: "《智能感知技术及在电气工程中的应用》", 31 May 2017 * |
| 郑红波 等: "方向场和频率场约束的结构光条纹图像增强算法", 《计算机科学》 * |
Cited By (16)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110414595B (en) * | 2019-07-25 | 2022-04-08 | 广西科技大学 | Method for estimating direction field of texture image with direction consistency |
| CN110414595A (en) * | 2019-07-25 | 2019-11-05 | 广西科技大学 | The orientation estimate method of texture image with orientation consistency |
| CN111353992A (en) * | 2020-03-10 | 2020-06-30 | 塔里木大学 | Agricultural product defect detection method and system based on textural features |
| CN111353992B (en) * | 2020-03-10 | 2023-04-07 | 塔里木大学 | Agricultural product defect detection method and system based on textural features |
| CN111539939A (en) * | 2020-04-27 | 2020-08-14 | 扬州哈工科创机器人研究院有限公司 | Defect detection method and device based on machine vision |
| CN111862080A (en) * | 2020-07-31 | 2020-10-30 | 易思维（杭州）科技有限公司 | Deep learning defect identification method based on multi-feature fusion |
| CN112200789B (en) * | 2020-10-16 | 2023-11-21 | 中国铁道科学研究院集团有限公司 | Image recognition method and device, electronic equipment and storage medium |
| CN112200789A (en) * | 2020-10-16 | 2021-01-08 | 中国铁道科学研究院集团有限公司 | Image identification method and device, electronic equipment and storage medium |
| CN113689427A (en) * | 2021-10-25 | 2021-11-23 | 常州微亿智造科技有限公司 | Defect detection method for parameter optimization based on space particle automatic attraction algorithm |
| CN115187590A (en) * | 2022-09-08 | 2022-10-14 | 山东艾克赛尔机械制造有限公司 | Automobile part defect detection method based on machine vision |
| CN115187590B (en) * | 2022-09-08 | 2022-12-20 | 山东艾克赛尔机械制造有限公司 | Automobile part defect detection method based on machine vision |
| CN115619767A (en) * | 2022-11-09 | 2023-01-17 | 南京云创大数据科技股份有限公司 | Method and device for detecting surface defects of mirror-like workpiece based on multi-illumination condition |
| CN115619767B (en) * | 2022-11-09 | 2023-04-18 | 南京云创大数据科技股份有限公司 | Method and device for detecting surface defects of mirror-like workpiece based on multi-illumination condition |
| CN117237340A (en) * | 2023-11-10 | 2023-12-15 | 江西省中鼐科技服务有限公司 | Method and system for detecting appearance of mobile phone shell based on artificial intelligence |
| CN117314914A (en) * | 2023-11-29 | 2023-12-29 | 广州市市政工程试验检测有限公司 | Defect identification method for engineering nondestructive testing image and related equipment |
| CN117314914B (en) * | 2023-11-29 | 2024-03-29 | 广州市市政工程试验检测有限公司 | Defect identification method for engineering nondestructive testing image and related equipment |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109829906A (en) | It is a kind of based on the workpiece, defect of the field of direction and textural characteristics detection and classification method | |
| CN114723701B (en) | Gear defect detection method and system based on computer vision | |
| US7760912B2 (en) | Image segregation system with method for handling textures | |
| US20060018539A1 (en) | Image processing method, image processing apparatus, and image processing program | |
| CN110930390B (en) | Chip pin missing detection method based on semi-supervised deep learning | |
| US8175390B2 (en) | System and method for illumination invariant image segmentation | |
| US9053537B2 (en) | Classifier for use in generating a diffuse image | |
| Ren et al. | Specular reflection separation with color-lines constraint | |
| CN109685760B (en) | MATLAB-based SLM powder bed powder laying image convex hull depression defect detection method | |
| WO2010065086A1 (en) | A constraint generator for use in image segregation | |
| EP2263191A1 (en) | System and method for identifying complex tokens in an image | |
| US20130342694A1 (en) | Method and system for use of intrinsic images in an automotive driver-vehicle-assistance device | |
| US8559714B2 (en) | Post processing for improved generation of intrinsic images | |
| CN104597057A (en) | Columnar diode surface defect detection device based on machine vision | |
| US8249342B1 (en) | Color analytics for a digital image | |
| CN105865329A (en) | Vision-based acquisition system for end surface center coordinates of bundles of round steel and acquisition method thereof | |
| CN110866915A (en) | Circular inkstone quality detection method based on metric learning | |
| US20140294296A1 (en) | Spatially varying log-chromaticity normals for use in an image process | |
| US8913829B2 (en) | Automatic processing scale estimation for use in an image process | |
| CN113252103A (en) | Method for calculating volume and mass of material pile based on MATLAB image recognition technology | |
| US8934735B2 (en) | Oriented, spatio-spectral illumination constraints for use in an image progress | |
| CN109461123B (en) | A kind of geometric distortion automatic analysis method for picture quality objective evaluating point diagram | |
| US20130114905A1 (en) | Post processing for improved generation of intrinsic images | |
| US8553979B2 (en) | Post processing for improved generation of intrinsic images | |
| CN115184362A (en) | Rapid defect detection method based on structured light projection |
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 | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20190531 |