CN117911368A - Coating surface defect detection method and system based on neural network model - Google Patents
Coating surface defect detection method and system based on neural network model Download PDFInfo
- Publication number
- CN117911368A CN117911368A CN202410078401.2A CN202410078401A CN117911368A CN 117911368 A CN117911368 A CN 117911368A CN 202410078401 A CN202410078401 A CN 202410078401A CN 117911368 A CN117911368 A CN 117911368A
- Authority
- CN
- China
- Prior art keywords
- surface image
- coating surface
- pixel
- image
- initial defect
- 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
Classifications
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; 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 OR CALCULATING; 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
- G06N3/0464—Convolutional networks [CNN, ConvNet]
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/13—Edge detection
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/40—Analysis of texture
- G06T7/41—Analysis of texture based on statistical description of texture
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10004—Still image; Photographic image
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- General Health & Medical Sciences (AREA)
- Artificial Intelligence (AREA)
- Computational Linguistics (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- Biomedical Technology (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- Biophysics (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Life Sciences & Earth Sciences (AREA)
- Probability & Statistics with Applications (AREA)
- Health & Medical Sciences (AREA)
- Quality & Reliability (AREA)
- Image Analysis (AREA)
Abstract
The invention provides a painting surface defect detection method and system based on a neural network model, comprising the steps of obtaining a painting surface image of an object to be detected, preprocessing the painting surface image, extracting texture features and edge features of the preprocessed painting surface image, performing traversal detection on pixels of the preprocessed painting surface image through an attention model to obtain an initial defect position, performing pixel assignment on a pixel area with a preset size where the initial defect position is located to obtain an initial defect model of the painting surface image, screening the initial defect model based on characteristic information, and determining the final defect position of the painting surface image; according to the method, the neural network model is adopted to extract the characteristic information of the painting surface image, the attention model is introduced to construct an initial defect model of the painting surface image, the initial defect model is screened based on the characteristic information, the accurate defect position can be obtained, and the precision of painting surface defect detection is improved.
Description
Technical Field
The invention relates to the technical field of coating surface defect detection, in particular to a coating surface defect detection method and system based on a neural network model.
Background
In conventional industrial production, especially in the paint industry, detection of surface defects often relies on manual visual inspection. The main disadvantages of this method are its inefficiency and the susceptibility of the test results to subjectivity of the test personnel. The accuracy and repeatability of manual detection are poor, and especially after continuous long-time work, detection personnel are easy to fatigue, so that judgment is affected. Furthermore, for small or complex defects, the human eye may not be able to accurately identify, resulting in missed detection.
Or defect detection is performed by means of image recognition, there is also a problem in that the detection accuracy is not high.
Disclosure of Invention
The invention provides a method and a system for detecting defects of a coating surface based on a neural network model, aiming at improving the accuracy of the detection of the defects of the coating surface.
In a first aspect, the present invention provides a method for detecting a defect on a coated surface based on a neural network model, including:
Acquiring a coating surface image of an object to be tested, and preprocessing the coating surface image;
extracting characteristic information of the pretreated coating surface image, wherein the characteristic information comprises texture characteristics and edge characteristics;
Traversing and detecting pixels of the pretreated coating surface image through an attention model to obtain an initial defect position, and carrying out pixel assignment on a pixel area with a preset size where the initial defect position is located based on pixel values of the original coating surface image at the defect position to obtain an initial defect model of the coating surface image;
And screening the initial defect model based on the characteristic information to determine the final defect position of the coating surface image.
In one embodiment, the acquiring a coating surface image of the object to be measured, and the preprocessing the coating surface image includes:
acquiring a coating surface image of an object to be measured, wherein the coating surface image comprises a light source area and a non-light source area;
Carrying out graying and binarization treatment on the coating surface image to obtain a black-and-white image of the coating surface image, and finding out a white area of the black-and-white image;
and expanding and dividing the white area to obtain a plurality of image units with preset sizes and containing the white area.
In one embodiment, performing traversal detection on pixels of the preprocessed coating surface image through an attention model to obtain initial defect positions, performing pixel assignment on a pixel area with a preset size where the initial defect positions are located, and obtaining an initial defect model of the coating surface image includes:
Performing traversal detection on the pixels of the pretreated coating surface image;
if the color of the pixels with the inconsistent colors is inconsistent with the peripheral color of the pixels with the inconsistent colors continuously, determining the pixel areas with the inconsistent colors as initial defect positions;
Determining a central pixel of the initial defect position, acquiring a pixel value of an original coating surface image in a pixel area with a preset size where the central pixel is located, and calculating an average value;
Subtracting the average value from all the pixel values in the pixel region with the preset size, carrying out normalization processing, and assigning the pixels in other regions of the original painting surface image to 0 to obtain an initial defect model of the painting surface image.
In one embodiment, the pixel area of the preset size is a 15×15 pixel area.
In one embodiment, the extracting feature information of the preprocessed painting surface image, the feature information including texture features and edge features includes:
Extracting basic texture features and edge features of the painting surface image through a convolution layer and an activation function;
finer texture features and edge features of the painted surface image are extracted by deep convolutional layers and residual blocks.
In one embodiment, after extracting the finer texture features and edge features of the painted surface image by the deep convolution layer and the residual block, the method further includes:
and integrating the basic texture features and the edge features and the finer texture features and the edge features, and refining the feature information of the painting surface image.
In one embodiment, the graying and binarizing the coated surface image to obtain a black-and-white image of the coated surface image, and finding a white area of the black-and-white image includes:
Carrying out graying and binarization treatment on the coating surface image, assigning zero to pixels of the non-light source area, and using a white area as a light source area;
And carrying out graying and binarization processing on the coating surface image, assigning the pixels of the light source area to be zero, and enabling the white area to be a non-light source area.
In a second aspect, the present invention provides a coating surface defect detection system based on a neural network model, including:
The pretreatment unit is used for acquiring a coating surface image of an object to be tested and carrying out pretreatment on the coating surface image;
the feature extraction unit is used for extracting feature information of the pretreated coating surface image, wherein the feature information comprises texture features and edge features;
the attention unit is used for performing traversal detection on the pixels of the preprocessed coating surface image through the attention model to obtain initial defect positions, and performing pixel assignment on a pixel area with a preset size where the initial defect positions are located to obtain an initial defect model of the coating surface image;
And the screening unit is used for screening the initial defect model based on the characteristic information and determining the final defect position of the coating surface image.
In a third aspect, the invention proposes a computer device comprising a processor and a memory, said memory having stored therein a computer program which, when loaded and executed by said processor, implements the method of any of the preceding claims.
In a fourth aspect, the invention proposes a computer storage medium having stored therein a computer program which, when executed, implements a method as claimed in any of the preceding claims.
The invention relates to a method and a system for detecting the defects of a coating surface based on a neural network model, which comprise the steps of obtaining a coating surface image of an object to be detected, preprocessing the coating surface image, extracting texture features and edge features of the preprocessed coating surface image, performing traversal detection on pixels of the preprocessed coating surface image through an attention model to obtain initial defect positions, performing pixel assignment on a pixel area with a preset size where the initial defect positions are located to obtain an initial defect model of the coating surface image, screening the initial defect model based on characteristic information, and determining final defect positions of the coating surface image; according to the method, the neural network model is adopted to extract the characteristic information of the painting surface image, the attention model is introduced to construct an initial defect model of the painting surface image, the initial defect model is screened based on the characteristic information, the accurate defect position can be obtained, and the precision of painting surface defect detection is improved.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required in the embodiments are briefly described below, and the drawings in the following description are only drawings corresponding to some embodiments of the present invention, and it is possible for a person skilled in the art to obtain drawings of other embodiments according to these drawings without inventive effort.
FIG. 1 is a flow chart of a method for detecting defects of a coated surface based on a neural network model according to one embodiment of the invention;
FIG. 2 is a schematic illustration of a coated surface image in accordance with one embodiment of the present invention;
FIG. 3 is a flow chart of a method for detecting defects of a coated surface based on a neural network model according to one embodiment of the invention;
FIG. 4 is a schematic diagram of an initial defect location according to one embodiment of the present invention;
fig. 5 is a schematic structural diagram of a coated surface defect detecting device based on a neural network model according to one embodiment of the present invention.
Detailed Description
The following description of the embodiments of the present invention will be made clearly and completely with reference to the accompanying drawings, in which it is apparent that the embodiments described are only some embodiments of the present invention, but not all embodiments. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.
Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs.
Referring to fig. 1, in one embodiment, a method for detecting a defect of a coated surface based on a neural network model according to the present invention includes:
Step S101, a coating surface image of an object to be detected is obtained, and the coating surface image is preprocessed.
Specifically, referring to fig. 2, a coated surface image of an object to be measured is obtained, the coated surface image including a light source region 2-1 and a non-light source region 2-2;
Carrying out graying and binarization treatment on the coating surface image to obtain a black-and-white image of the coating surface image, and finding out a white area of the black-and-white image;
And expanding and dividing the white area to obtain a plurality of image units 2-3 with preset sizes and containing the white area.
When the light of the detection light source is captured by the camera after being reflected by the detected object, light spots similar to the light source can be formed in the image, for example, when the light source is an LED tube, the image obtained by the camera forms stripes with alternate brightness, and the stripes with alternate black and white are obtained after the graying and binarization treatment, so that the defects are more obvious.
The painting surface image is standardized through preprocessing, including color conversion, size adjustment and normalization, and the original image data is converted into a format and size which can be effectively processed by the neural network model.
And step S102, extracting characteristic information of the preprocessed coating surface image, wherein the characteristic information comprises texture characteristics and edge characteristics.
Specifically, extracting basic texture features and edge features of the painting surface image from each image unit through a convolution layer and an activation function;
Finer texture features and edge features of the painted surface image are extracted by deep convolutional layers and residual blocks. The integrity of the features is maintained by the residual block connection, preventing the problem of gradient vanishing in deep networks.
Step S103, performing traversal detection on the pixels of the pretreated coating surface image through the attention model to obtain an initial defect position, and performing pixel assignment on a pixel area with a preset size where the initial defect position is located based on the pixel value of the original coating surface image at the defect position to obtain an initial defect model of the coating surface image.
The pixels of each image unit are subjected to traversal detection, and the defect positions, such as abnormal pixel blocks, are preliminarily determined according to the characteristics of the pixels. And carrying out pixel assignment on a pixel area with a preset size where the preliminarily determined defect position is based on the pixel value of the original coating surface image at the defect position, so as to obtain an initial defect model of the coating surface image. Specifically, for example, the pixel value of the initial defect position is reserved, and the other area is assigned 0.
And step S104, screening the initial defect model based on the characteristic information, and determining the final defect position of the coating surface image.
The texture features and the edge features of the painting surface image comprise feature information of parts such as normal areas, defects and noise of the painting surface image, and the initial defect model is screened based on the extracted texture features and the edge features of the painting surface image, so that the noise in the initial defect model can be filtered, and the accurate defect position can be obtained.
According to the method for detecting the defects of the coated surface based on the neural network model, the coated surface image is preprocessed to obtain a black-and-white image unit, so that defects are more obvious, the neural network model is adopted to extract texture features and edge features of the coated surface image, defect positions are preliminarily determined according to the attention model, pixel assignment is carried out on pixel regions with preset sizes of the defect positions, an initial defect model of the coated surface image is obtained, the initial defect model is screened based on the texture features and the edge features of the coated surface image, accurate defect positions can be obtained, noise is filtered, and the accuracy of detecting the defects of the coated surface is improved.
Referring to fig. 3, in one embodiment, a method for detecting a defect of a coated surface based on a neural network model according to the present invention includes:
step S201, a coated surface image of the object to be measured is acquired, wherein the coated surface image includes a light source region and a non-light source region.
Step S202, carrying out gray scale and binarization processing on the coating surface image to obtain a black-and-white image of the coating surface image, and finding out a white area of the black-and-white image.
In one embodiment, the graying and binarizing the coated surface image to obtain a black-and-white image of the coated surface image, and finding a white area of the black-and-white image includes:
Carrying out graying and binarization treatment on the coating surface image, assigning zero to pixels of the non-light source area, and using a white area as a light source area;
And carrying out graying and binarization processing on the coating surface image, assigning the pixels of the light source area to be zero, and enabling the white area to be a non-light source area.
The white area and the defects around the white area are more obvious after binarization of the coating surface image, in order to accurately detect the defects of the whole visual field of the image, firstly, gray level and binarization processing is carried out on the acquired coating surface image, at the moment, the binarized white area corresponds to a light source area of the coating surface image, the black area corresponds to a non-light source area of the coating surface image, and the defect detection is carried out on the coating surface image after the processing; and then carrying out grey-scale and binarization processing on the acquired coating surface image again, at the moment, assigning the pixels of the light source area to be zero, and continuously carrying out defect detection on the coating surface image after the processing on the white area corresponding to the non-light source area of the coating surface image, so that the defect detection precision of the whole visual field of the coating surface image can be ensured.
Step S203, performing expansion and segmentation based on the white area, to obtain a plurality of image units with preset sizes including the white area.
And S204, extracting basic texture features and edge features of the painting surface image through a convolution layer and an activation function.
Step S205, extracting finer texture features and edge features of the painted surface image through the deep convolution layer and the residual block.
And S206, integrating the basic texture features and the edge features and the finer texture features and the edge features, and refining the feature information of the painting surface image.
The integration and refinement of the feature information includes full connection layer, batch normalization, dropout to enhance the generalization ability of the neural network model in preparation for final classification or regression tasks.
Step S207, performing traversal detection on the pixels of the preprocessed coated surface image.
In step S208, if the preset number of pixel colors are inconsistent with the peripheral colors, determining the pixel area with inconsistent colors as the initial defect position.
And (3) carrying out gray value exploration on each segmented image unit, if a preset number of pixel colors are inconsistent with the peripheral colors, such as 5 pixel points, determining the pixel areas with inconsistent colors as initial defect positions, wherein the pixel areas are not limited to the 5 pixel points, and the embodiment does not limit the initial defect positions.
Step S209, determining a center pixel of the initial defect position, obtaining a pixel value of an original painting surface image in a pixel area with a preset size where the center pixel is located, calculating an average value, subtracting the average value from all pixel values in the pixel area with the preset size, performing normalization processing, and assigning pixels in other areas of the original painting surface image to 0, so as to obtain an initial defect model of the painting surface image.
Referring to fig. 4, a black font area 4-1 is a detected initial defect position, a pixel value of a central pixel of the initial defect position is 131, an average value in a 15x15 pixel area where the initial defect position is located is calculated, all pixel values of the area are subtracted by the average value, a new pixel map can be obtained, normalization processing is performed on the pixel map, a pixel value range is scaled to be between 0 and 1, pixels in other areas are assigned to be 0, and an initial defect model is obtained, wherein a 15x15 pixel area is a preferred area.
In one embodiment, the method further comprises the steps of multiplying the feature map corresponding to the texture features and the edge features of the painting surface image by the initial defect model after coordinate matching, so that the difference value between the true defect with smaller feature difference and noise can be enlarged, the multiplied result is trained again, the trained feature model can describe the features of the true defect more comprehensively and accurately, and noise in the initial defect model can be filtered more effectively based on the feature model.
And step S210, screening the initial defect model based on the characteristic information, and determining the final defect position of the coating surface image.
After texture features and edge features of the painting surface image are obtained, defect positions initially determined by the initial defect model can be screened based on the feature information, noise is filtered, and more accurate defect positions are obtained.
According to the method for detecting the defects of the coating surface based on the neural network model, accurate cultural characteristics and edge characteristics of the coating surface image are extracted through the network model, the attention model is constructed to preliminarily determine the defect positions, the pixels of the defect positions are assigned to obtain an initial defect model, the initial defect model is screened and filtered according to the accurate cultural characteristics and the edge characteristics of the coating surface image, the accurate defect positions can be obtained, noise is filtered, and the accuracy of detecting the defects of the coating surface is improved.
Referring to fig. 5, an embodiment of the present application further provides a coating surface defect detection system based on a neural network model, in one embodiment, the system includes:
a preprocessing unit 10, configured to acquire a coating surface image of an object to be measured, and perform preprocessing on the coating surface image.
And a feature extraction unit 20 for extracting feature information of the pretreated painting surface image, the feature information including texture features and edge features.
The attention unit 30 is configured to perform traversal detection on the pixel of the preprocessed coating surface image through the attention model to obtain an initial defect position, and perform pixel assignment on a pixel area with a preset size where the initial defect position is located based on a pixel value of the original coating surface image at the defect position to obtain an initial defect model of the coating surface image.
And a screening unit 40, configured to screen the initial defect model based on the feature information, and determine a final defect position of the coated surface image.
In one embodiment, the preprocessing unit 10 is specifically configured to acquire a coated surface image of an object to be measured, where the coated surface image includes a light source area and a non-light source area;
Carrying out graying and binarization treatment on the coating surface image to obtain a black-and-white image of the coating surface image, and finding out a white area of the black-and-white image;
and expanding and dividing the white area to obtain a plurality of image units with preset sizes and containing the white area.
In one embodiment, the graying and binarizing the coated surface image to obtain a black-and-white image of the coated surface image, and finding a white area of the black-and-white image includes:
carrying out graying and binarization treatment on the coating surface image, assigning zero to pixels of the light source area, wherein the white area is the light source area;
And carrying out graying and binarization processing on the coating surface image, assigning the pixels of the non-light source area to be zero, and enabling the white area to be the non-light source area.
In one embodiment, the attention unit 30 is specifically configured to:
Performing traversal detection on the pixels of the pretreated coating surface image;
if the color of the pixels with the inconsistent colors is inconsistent with the peripheral color of the pixels with the inconsistent colors continuously, determining the pixel areas with the inconsistent colors as initial defect positions;
Determining a central pixel of the initial defect position, acquiring a pixel value of an original coating surface image in a pixel area with a preset size where the central pixel is located, and calculating an average value;
Subtracting the average value from all the pixel values in the pixel region with the preset size, carrying out normalization processing, and assigning the pixels in other regions of the original painting surface image to 0 to obtain an initial defect model of the painting surface image.
In one embodiment, the feature extraction unit 20 is specifically configured to:
Extracting basic texture features and edge features of the painting surface image through a convolution layer and an activation function;
finer texture features and edge features of the painted surface image are extracted by deep convolutional layers and residual blocks.
In one embodiment, the neural network model-based paint surface defect detection system further comprises:
And the integration unit is used for integrating the basic texture features and the edge features and finer texture features and edge features and refining the feature information of the coating surface image.
The specific process of executing the corresponding steps by each unit is described in detail in the above method embodiment, and is not described herein for brevity.
The embodiment of the invention also provides a computer device, which comprises a processor and a memory, wherein the memory stores a computer program, and the computer program realizes the steps of any one of the above method embodiments when being loaded and executed by the processor.
The embodiment of the invention also provides a computer storage medium, wherein a computer program is stored in the computer storage medium, and when the computer program is executed, the method steps of any one of the above method embodiments are realized.
In the foregoing embodiments of the present invention, it should be understood that the disclosed systems, devices and methods may be implemented in other manners. For example, the apparatus embodiments described above are merely illustrative, e.g., the division of the units is merely a logical function division, and there may be additional divisions when actually implemented, e.g., multiple units or components may be combined or integrated into another system, or some features may be omitted or not performed. Alternatively, the coupling or direct coupling or communication connection shown or discussed with each other may be an indirect coupling or communication connection via some interfaces, devices or units, which may be in electrical, mechanical or other form.
The units described as separate units may or may not be physically separate, and units shown as units may or may not be physical units, may be located in one place, or may be distributed on a plurality of network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.
In addition, each functional unit in the embodiments of the present application may be integrated in one processing unit, or each unit may exist alone physically, or two or more units may be integrated in one unit. The integrated units may be implemented in hardware or in software functional units. The integrated units, if implemented in the form of software functional units and sold or used as stand-alone products, may be stored in a computer readable storage medium.
Based on such understanding, the technical solution of the present application may be embodied in essence or contributing part or all or part of the technical solution in the form of a software product stored in a storage medium, comprising several instructions for causing a computer device (which may be a mobile terminal, a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method according to the embodiments of the present application. And the aforementioned storage medium includes: a usb disk, a removable hard disk, a read-only memory (ROM), a random access memory (RAM, random Access Memory), a magnetic disk, or an optical disk, or other various media capable of storing program codes.
In summary, although the present invention has been described with reference to the preferred embodiments, the scope of the invention is not limited thereto, and any person skilled in the art who is skilled in the art should make equivalent substitutions or modifications according to the technical scheme of the present invention within the scope of the present invention.
The technical features of the above-described embodiments may be arbitrarily combined, and all possible combinations of the technical features in the above-described embodiments are not described for brevity of description, however, as long as there is no contradiction between the combinations of the technical features, they should be considered as the scope of the description.
Claims (10)
1. The coating surface defect detection method based on the neural network model is characterized by comprising the following steps of:
Acquiring a coating surface image of an object to be tested, and preprocessing the coating surface image;
extracting characteristic information of the pretreated coating surface image, wherein the characteristic information comprises texture characteristics and edge characteristics;
Traversing and detecting pixels of the pretreated coating surface image through an attention model to obtain an initial defect position, and carrying out pixel assignment on a pixel area with a preset size where the initial defect position is located based on pixel values of the original coating surface image at the defect position to obtain an initial defect model of the coating surface image;
And screening the initial defect model based on the characteristic information to determine the final defect position of the coating surface image.
2. The method of claim 1, wherein the acquiring a painted surface image of the object to be measured, the preprocessing of the painted surface image comprising:
acquiring a coating surface image of an object to be measured, wherein the coating surface image comprises a light source area and a non-light source area;
Carrying out graying and binarization treatment on the coating surface image to obtain a black-and-white image of the coating surface image, and finding out a white area of the black-and-white image;
and expanding and dividing the white area to obtain a plurality of image units with preset sizes and containing the white area.
3. The method of claim 1, wherein performing traversal detection on the pixels of the preprocessed paint surface image through an attention model to obtain an initial defect position, performing pixel assignment on a pixel area with a preset size where the initial defect position is located based on pixel values of the original paint surface image at the defect position, and obtaining an initial defect model of the paint surface image includes:
Performing traversal detection on the pixels of the pretreated coating surface image;
if the color of the pixels with the inconsistent colors is inconsistent with the peripheral color of the pixels with the inconsistent colors continuously, determining the pixel areas with the inconsistent colors as initial defect positions;
Determining a central pixel of the initial defect position, acquiring a pixel value of an original coating surface image in a pixel area with a preset size where the central pixel is located, and calculating an average value;
Subtracting the average value from all the pixel values in the pixel region with the preset size, carrying out normalization processing, and assigning the pixels in other regions of the original painting surface image to 0 to obtain an initial defect model of the painting surface image.
4. A method according to claim 3, wherein the predetermined size of pixel area is 15 x 15 pixel area.
5. The method of claim 1, wherein extracting feature information of the preprocessed paint surface image, the feature information including texture features and edge features, comprises:
Extracting basic texture features and edge features of the painting surface image through a convolution layer and an activation function;
finer texture features and edge features of the painted surface image are extracted by deep convolutional layers and residual blocks.
6. The method of claim 5, wherein after extracting finer texture features and edge features of the painted surface image by deep convolutional layers and residual blocks, further comprising:
and integrating the basic texture features and the edge features and the finer texture features and the edge features, and refining the feature information of the painting surface image.
7. The method of claim 2, wherein the graying and binarizing the painted surface image to obtain a black-and-white image of the painted surface image, and finding a white area of the black-and-white image comprises:
Carrying out graying and binarization treatment on the coating surface image, assigning zero to pixels of the non-light source area, and using a white area as a light source area;
And carrying out graying and binarization processing on the coating surface image, assigning the pixels of the light source area to be zero, and enabling the white area to be a non-light source area.
8. A paint surface defect detection system based on a neural network model, comprising:
The pretreatment unit is used for acquiring a coating surface image of an object to be tested and carrying out pretreatment on the coating surface image;
the feature extraction unit is used for extracting feature information of the pretreated coating surface image, wherein the feature information comprises texture features and edge features;
the attention unit is used for performing traversal detection on the pixels of the preprocessed coating surface image through the attention model to obtain initial defect positions, and performing pixel assignment on a pixel area with a preset size where the initial defect positions are located to obtain an initial defect model of the coating surface image;
And the screening unit is used for screening the initial defect model based on the characteristic information and determining the final defect position of the coating surface image.
9. A computer device comprising a processor and a memory, the memory having stored therein a computer program which, when loaded and executed by the processor, implements the method of any of claims 1 to 7.
10. A computer storage medium, characterized in that the computer storage medium has stored therein a computer program which, when executed, implements the method according to any of claims 1 to 7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202410078401.2A CN117911368A (en) | 2024-01-18 | 2024-01-18 | Coating surface defect detection method and system based on neural network model |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202410078401.2A CN117911368A (en) | 2024-01-18 | 2024-01-18 | Coating surface defect detection method and system based on neural network model |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN117911368A true CN117911368A (en) | 2024-04-19 |
Family
ID=90694152
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202410078401.2A Pending CN117911368A (en) | 2024-01-18 | 2024-01-18 | Coating surface defect detection method and system based on neural network model |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117911368A (en) |
-
2024
- 2024-01-18 CN CN202410078401.2A patent/CN117911368A/en active Pending
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN115222733B (en) | Metal component surface defect detection method based on image recognition | |
| CN115908269B (en) | Visual defect detection method, visual defect detection device, storage medium and computer equipment | |
| CN113570605B (en) | Defect detection method and system based on liquid crystal display panel | |
| CN105139386B (en) | A kind of image processing method of fast automatic detecting electric connector solder joint defective work | |
| CN117392042A (en) | Defect detection method, defect detection equipment and storage medium | |
| CN115205223B (en) | Visual inspection method and device for transparent object, computer equipment and medium | |
| CN114937055A (en) | Image self-adaptive segmentation method and system based on artificial intelligence | |
| CN114140679A (en) | Defect fusion method, device, recognition system and storage medium | |
| CN116152242B (en) | A Visual Detection System for Natural Leather Defects Used in Basketball | |
| CN118279304B (en) | Abnormal recognition method, device and medium for special-shaped metal piece based on image processing | |
| CN114202532A (en) | Spraying defect detection method, device, equipment and storage medium | |
| CN115526889B (en) | Nondestructive testing method of boiler pressure pipeline based on image processing | |
| CN120147263B (en) | An automatic bearing surface defect detection system based on image recognition | |
| CN118501177A (en) | Appearance defect detection method and system for formed foil | |
| CN116883408A (en) | Integrating instrument shell defect detection method based on artificial intelligence | |
| JP7383722B2 (en) | Method and system for defect detection in image data of target coatings | |
| CN119991544A (en) | Intelligent classification method and system for transmission line tension clamp defects based on two-dimensional images | |
| CN112950594A (en) | Method and device for detecting surface defects of product and storage medium | |
| CN111738984A (en) | Method and system for skin image blob evaluation based on watershed and seed filling | |
| CN117911368A (en) | Coating surface defect detection method and system based on neural network model | |
| CN116703912B (en) | Mini-host network port integrity visual detection method | |
| CN119714800B (en) | A high-magnification light source uniformity detection method and system | |
| CN121033022B (en) | A method for detecting abnormal defects in terminal welding | |
| Rajitha et al. | A new local homogeneity analysis method based on pixel intensities for image defect detection | |
| CN121437915A (en) | A computer vision-based method and system for detecting colloidal gold |
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 |