Coating thickness detection system and method based on visual monitoring

Abstract

The invention discloses a coating thickness detection system and method based on visual monitoring, and relates to the technical field of visual monitoring, and the method comprises the steps: carrying out the preliminary prediction of the coating thickness through geometrical optics; performing coating thickness prediction through a multi-channel neural network based on the pre-processed coating surface image and the preliminary prediction value; according to the predicted thickness of the coating, defect identification and evaluation are carried out, and a final coating thickness and defect comprehensive evaluation result is generated; and real-time feedback is carried out through a flaw comprehensive evaluation result. According to the method, the coating surface image is collected in real time and preprocessed, primary thickness prediction is carried out in combination with geometrical optics, accurate coating thickness prediction is carried out through the multi-channel neural network, and the prediction precision and the adaptive capacity are remarkably improved. Through the defect identification and evaluation module, the type, the shape, the complexity and the position of the coating surface defect are comprehensively considered, and the coating defect is accurately identified and quantitatively evaluated, so that the accuracy and the robustness of defect detection are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of visual monitoring, and in particular to a coating thickness detection system and method based on visual monitoring. Background Art

[0002] With the continuous advancement of industrialization, coating technology has been widely used in many fields, such as automobile manufacturing, aerospace, electronic equipment, construction, etc. Coatings can not only improve the appearance of products, but also provide various functions such as corrosion resistance, oxidation resistance, and wear resistance. Therefore, the quality of the coating directly affects the performance and service life of the product. In order to ensure the uniformity and quality of the coating, accurate measurement of coating thickness has become a vital task. Traditional coating thickness detection methods mainly rely on physical contact measurement tools, such as electromagnetic thickness gauges, eddy current flaw detectors, and ultrasonic thickness gauges. Although these methods can provide more accurate thickness data, they usually require contact on the coating surface, which may cause damage to the coating surface, and are not suitable for complex curved surfaces or places that are difficult to contact.

[0003] In recent years, with the rapid development of image processing technology, machine learning and deep learning, the coating thickness detection method based on visual monitoring has become an important research direction. By collecting images of the coating surface with a high-resolution camera and analyzing the coating with an image processing algorithm, non-contact measurement can be achieved, which not only avoids the limitations of traditional methods, but also can efficiently perform real-time detection on large-scale production lines. In addition, by combining geometric optical models with image processing technology, the relationship between the intensity of reflected light and the angle of incident light is used to make a preliminary prediction of the coating thickness, further improving the accuracy of detection. However, there are two main deficiencies in the existing technology: first, most image-based thickness prediction methods rely too much on manually set parameters and rules and lack adaptability; second, existing defect detection methods usually rely on a single image feature extraction method, which cannot fully utilize the complex information of the coating surface, resulting in insufficient accuracy and robustness of defect recognition. Summary of the invention

[0004] In view of the above existing problems, the present invention is proposed.

[0005] Therefore, the present invention provides a coating thickness detection method based on visual monitoring to solve the problem that the image thickness prediction method lacks adaptability and the existing defect detection method cannot fully utilize the complex information of the coating surface.

[0006] In order to solve the above technical problems, the present invention provides the following technical solutions:

[0007] In a first aspect, the present invention provides a coating thickness detection method based on visual monitoring, which includes real-time acquisition of coating surface images, and preprocessing of the acquired coating surface images; preliminary prediction of coating thickness through geometric optics; based on the preprocessed images and preliminary prediction values, coating thickness prediction through a multi-channel neural network; defect identification and evaluation based on the predicted coating thickness to generate a final coating thickness and defect comprehensive evaluation result; and real-time feedback based on the defect comprehensive evaluation result.

[0008] As a preferred solution of the coating thickness detection method based on visual monitoring of the present invention, the specific steps of real-time acquisition of coating surface image are as follows:

[0009] The coating surface image is captured in real time using a high-resolution camera.

[0010] As a preferred solution of the coating thickness detection method based on visual monitoring of the present invention, the preprocessing of the collected image of the coating surface is carried out in the following specific steps:

[0011] The collected images of the coating surface are subjected to denoising, image enhancement, illumination equalization, image segmentation, image alignment and geometric correction.

[0012] As a preferred solution of the coating thickness detection method based on visual monitoring of the present invention, the coating thickness is initially predicted by geometric optics, and the specific steps are as follows:

[0013] Select the reflection method as the calculation method, use the optical reflection coefficient of the coating and the incident light angle to calculate the thickness, receive the reflected light intensity on the coating surface, measure the ratio of the reflected light intensity to the incident light intensity, and obtain the reflectivity;

[0014] Based on the principle of geometric optics, the coating thickness is preliminarily predicted as follows:

[0015]

[0016] Among them, H is the preliminary predicted coating thickness, λ is the wavelength of the light wave, and R is the reflectivity.

[0017] As a preferred solution of the coating thickness detection method based on visual monitoring of the present invention, the coating thickness prediction is performed based on the preprocessed image and the preliminary prediction value through a multi-channel neural network. The specific steps are as follows:

[0018] Based on the multi-channel neural network, the multi-channel neural network structure is divided into two independent channels. Channel 1 extracts the spatial features of the coating surface image through a deep convolutional neural network after preprocessing, and forms the feature vector of the image channel by splicing;

[0019] Channel 2 inputs the preliminary prediction value into the multilayer perceptron, performs nonlinear mapping through the fully connected layer, and generates the feature vector of the preliminary prediction value;

[0020] The feature vector of the image channel and the feature vector of the preliminary prediction value are weighted and combined through the weighted fusion layer to obtain the final fused feature vector. The final fused feature vector is used for the final coating thickness prediction through the fully connected layer to generate the predicted coating thickness.

[0021] As a preferred solution of the coating thickness detection method based on visual monitoring of the present invention, wherein: the defect identification and evaluation are performed according to the coating thickness predicted by the coating to generate the final coating thickness and defect evaluation results, the specific steps are as follows:

[0022] Extract edge features of the preprocessed coating surface image, detect circular and linear defects on the coating surface, and perform segmentation and defect recognition on the coating surface in combination with morphological operations;

[0023] The area and shape characteristics of the coating surface defect area are obtained through image analysis algorithms, and the complexity score of the coating surface defect is calculated based on the shape characteristics;

[0024] The relative position impact of each defect area is defined by calculating the distance between the defect area and the edge of the coating. The coating surface defect severity score is calculated based on the comprehensive score of the defect area, the complexity score of the defect and the relative position impact score. The expression is:

[0025]

[0026] Among them, E is the severity score of coating surface defects, A is n is the area of ​​the nth coating surface defect area, A max is the maximum area of ​​the coating surface, C s is the complexity score of coating surface defects, s indicates that the complexity score of coating surface defects is related to the complexity of the geometric shape of the defects, C l is the relative position impact score, l means the relative position impact score is related to the position relationship of the defect relative to the coating edge;

[0027] Based on the preliminary thickness predicted by geometric optics, the coating thickness predicted by the multi-channel neural network, and the coating surface defect severity score, the coating thickness is adjusted by a nonlinear function and the final adjusted coating thickness is calculated in combination with environmental factors. The expression is:

[0028]

[0029] Among them, H fis the final adjusted coating thickness, P is the coating thickness predicted by the multi-channel neural network, ΔH is the thickness error feedback, γ is the error correction coefficient, and α is the environmental correction coefficient;

[0030] Based on the final adjusted coating thickness and the standard thickness of the coating, a coating quality assessment result is generated.

[0031] As a preferred solution of the coating thickness detection method based on visual monitoring of the present invention, wherein: the comprehensive defect evaluation result is used to provide real-time feedback, and the specific steps are as follows:

[0032] Based on the comprehensive defect evaluation results, thresholds are set based on historical data. When the defect density and size are greater than the threshold, the coating quality is deemed unqualified. The operator is notified through a real-time alarm, and the coating thickness and defect score are recorded in real time.

[0033] In a second aspect, the present invention provides a coating thickness detection system based on visual monitoring, comprising an image acquisition and preprocessing module, a coating thickness preliminary prediction module, a coating thickness prediction module, a coating thickness and defect comprehensive evaluation module and a real-time feedback module.

[0034] The image acquisition and preprocessing module acquires the coating surface image in real time and preprocesses the acquired coating surface image; the coating thickness preliminary prediction module makes a preliminary prediction of the coating thickness through geometric optics; the coating thickness prediction module predicts the coating thickness through a multi-channel neural network based on the preprocessed image and the preliminary prediction value; the coating thickness and defect comprehensive evaluation module identifies and evaluates defects according to the predicted coating thickness to generate the final coating thickness and defect comprehensive evaluation results; the real-time feedback module provides real-time feedback through the defect comprehensive evaluation results.

[0035] In a third aspect, the present invention provides a computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: when the computer program is executed by the processor, any step of the coating thickness detection method based on visual monitoring as described in the first aspect of the present invention is implemented.

[0036] In a fourth aspect, the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein: when the computer program is executed by a processor, any step of the coating thickness detection method based on visual monitoring as described in the first aspect of the present invention is implemented.

[0037] The beneficial effects of the present invention are as follows: the present invention collects the coating surface image in real time and performs preprocessing, combines geometric optics to make preliminary thickness prediction, and then uses a multi-channel neural network to make accurate coating thickness prediction, which significantly improves the prediction accuracy and adaptability. Furthermore, through the defect recognition and evaluation module, the type, shape, complexity and location of the coating surface defects are comprehensively considered, and the coating defects are accurately identified and quantitatively evaluated, thereby improving the accuracy and robustness of defect detection. Finally, based on the comprehensive evaluation results, real-time feedback is provided, the operator is notified in time and relevant data is recorded to ensure that the coating quality is effectively controlled, optimize the quality management in the production process, and has broad application prospects. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative work.

[0039] Figure 1 This is a flow chart of the coating thickness detection method based on visual monitoring in Example 1.

[0040] Figure 2 Schematic diagram of the coating thickness detection system based on visual monitoring in Example 1. DETAILED DESCRIPTION

[0041] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are described in detail below in conjunction with the accompanying drawings.

[0042] In the following description, many specific details are set forth to facilitate a full understanding of the present invention, but the present invention may also be implemented in other ways different from those described herein, and those skilled in the art may make similar generalizations without violating the connotation of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.

[0043] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The term "in one embodiment" that appears in different places in this specification does not necessarily refer to the same embodiment, nor does it refer to a separate or selective embodiment that is mutually exclusive with other embodiments.

[0044] Example 1, reference Figure 1 and Figure 2 , which is the first embodiment of the present invention, provides a coating thickness detection method based on visual monitoring, comprising the following steps:

[0045] S1: real-time acquisition of coating surface images, and preprocessing of the acquired coating surface images;

[0046] Furthermore, through high-resolution cameras, images of the coating surface are collected in real time;

[0047] It should be noted that the high-resolution camera should be selected according to the characteristics of the coating and the application environment, and the light source should be selected according to the coating material, surface gloss and environmental conditions of the acquisition.

[0048] De-noising, image enhancement, illumination balance, image segmentation, image alignment and geometric correction are performed on the collected images of the coating surface;

[0049] It should be noted that denoising is performed by an image denoising algorithm, image enhancement is performed by histogram equalization and edge enhancement, illumination equalization is performed by a Retinex algorithm, image segmentation is performed by an image segmentation algorithm, and image alignment and geometric correction are performed by a geometric correction algorithm.

[0050] S2: Preliminary prediction of coating thickness by geometric optics;

[0051] Furthermore, the reflection method is selected as the calculation method, and the optical reflection coefficient of the coating and the incident light angle are used to estimate the thickness. The reflected light intensity is received on the coating surface, and the ratio of the reflected light intensity to the incident light intensity is measured to obtain the reflectivity, which is expressed as:

[0052]

[0053] Where R is the reflectivity, I r is the reflected light intensity, I0 is the incident light intensity;

[0054] Based on the principle of geometric optics, the coating thickness is preliminarily predicted as follows:

[0055]

[0056] Among them, H is the preliminary predicted coating thickness and λ is the wavelength of the light wave.

[0057] It should be noted that, according to the principles of geometric optics, it is assumed that the coating material is transparent and has a known refractive index, and the reflection and refraction behavior of the coating thickness at a specific wavelength can be described by the Fresnel formula. Considering the relationship between the thickness and the reflectivity of the coating, the interference effect is introduced to obtain a preliminary predicted coating thickness expression. According to the actually measured reflectivity and incident angle, the preliminary predicted coating thickness expression is used to predict the coating thickness.

[0058] S3: Based on the preprocessed image and preliminary prediction value, the coating thickness is predicted through a multi-channel neural network;

[0059] Furthermore, based on the multi-channel neural network, the multi-channel neural network structure is divided into two independent channels. Channel one is to extract the spatial features in the coating surface image through the deep convolutional neural network after the preprocessing, and form the feature vector of the image channel by splicing;

[0060] It should be noted that the preprocessed image is extracted through multi-scale convolution features, and the expression is:

[0061]

[0062] Among them, Y l is the convolution feature map, represents the Kth convolution kernel in the lth layer, * represents the convolution operation, K represents the number of multi-scale convolution kernels, and b represents the activation function;

[0063] Global average pooling is performed on the last layer of the convolutional feature map to reduce the feature dimension and preserve the global information;

[0064] Channel 2 inputs the preliminary prediction value into the multilayer perceptron, performs nonlinear mapping through the fully connected layer, and generates the feature vector of the preliminary prediction value;

[0065] Based on the preliminary predicted thickness, a multilayer perceptron is used to enhance the features of the preliminary predicted value. The expression is:

[0066] H (l) =f(U (l) H (l-1) +v (l) );

[0067] Among them, H (l) is the output of layer l, U (l) is the bias of the lth layer, H (l-1) is the input of the l-1th layer, v (l) is the activation function;

[0068] The feature vector of the image channel and the feature vector of the preliminary prediction value are weighted and combined through the weighted fusion layer to obtain the final fused feature vector. The final fused feature vector is used for the final coating thickness prediction through the fully connected layer to generate the predicted coating thickness.

[0069] It should be noted that the multilayer perceptron consists of three fully connected layers and one output layer;

[0070] The feature vector of the image channel and the feature vector of the preliminary prediction value are weighted and combined through the weighted fusion layer to obtain the final fused feature vector. The final fused feature vector is used for the final coating thickness prediction through the fully connected layer to generate the predicted coating thickness.

[0071] S4: Based on the predicted coating thickness, defect identification and evaluation are performed to generate the final coating thickness and defect comprehensive evaluation results;

[0072] Furthermore, the edge features of the preprocessed coating surface image are extracted by Sobel operator, and the circular and linear defects on the coating surface are detected by Hough transform. Based on the extracted edge features, the coating surface is segmented and defect recognition is performed in combination with morphological operations;

[0073] The area and shape characteristics of the coating surface defect area are obtained through the image analysis algorithm. The complexity score of the coating surface defect is calculated according to the shape characteristics. The relative position influence is defined by calculating the distance between each defect area and the coating edge. The coating surface defect severity score is calculated based on the comprehensive score of the area of ​​the coating surface defect area, the complexity score of the defect and the relative position influence score. The expression is:

[0074]

[0075] Among them, E is the severity score of coating surface defects, A is n is the area of ​​the nth coating surface defect area, A max is the maximum area of ​​the coating surface, C s is the complexity score of coating surface defects, s indicates that the complexity score of coating surface defects is related to the complexity of the geometric shape of the defects, C l is the relative position impact score, l means the relative position impact score is related to the position relationship of the defect relative to the coating edge;

[0076] Based on the preliminary thickness predicted by geometric optics, the coating thickness predicted by the multi-channel neural network, and the coating surface defect severity score, the coating thickness is adjusted by a nonlinear function and the final adjusted coating thickness is calculated in combination with environmental factors. The expression is:

[0077]

[0078] Among them, H f is the final adjusted coating thickness, P is the coating thickness predicted by the multi-channel neural network, ΔH is the thickness error feedback, γ is the error correction coefficient, α is the environmental correction coefficient, and F is the normalized parameter of environmental influence;

[0079] It should be noted that the thickness error feedback ΔH = H a -H s , H a The predicted coating thickness is combined with the preliminary predicted coating thickness, and the intermediate adjusted thickness H is generated by averaging several a , H s is the standard thickness of the coating;

[0080] The coating quality assessment result is generated based on the final adjusted coating thickness and the standard thickness of the coating. It should be noted that the standard thickness of the coating is an important benchmark value for measuring whether the coating quality is qualified, and different fields usually have clear industry standards or specifications for coating thickness.

[0081] It should be noted that the coating quality assessment result is the ratio of the final adjusted coating thickness to the standard coating thickness. When the ratio is equal to 1, it means that the coating thickness and quality are in full compliance with the standard. When the ratio is less than 1, it means that the coating thickness is lower than the standard, there are defects, and the quality is unqualified. When the ratio is greater than 1, it means that the coating thickness is higher than the standard, the coating is too thick, and the quality is unqualified.

[0082] S5: Provide real-time feedback through comprehensive defect evaluation results.

[0083] Furthermore, based on the comprehensive defect assessment results, when quality failure occurs, the operator is notified through a real-time alarm, and the coating thickness and defect score are recorded in real time.

[0084] This embodiment also provides a coating thickness detection system based on visual monitoring, comprising:

[0085] An image acquisition and preprocessing module acquires coating surface images in real time and preprocesses the acquired coating surface images;

[0086] The module for preliminary prediction of coating thickness makes a preliminary prediction of coating thickness through geometric optics;

[0087] The coating thickness prediction module predicts the coating thickness based on the preprocessed image and the preliminary prediction value through a multi-channel neural network;

[0088] The coating thickness and defect comprehensive evaluation module identifies and evaluates defects based on the predicted coating thickness and generates the final coating thickness and defect comprehensive evaluation results;

[0089] The real-time feedback module provides real-time feedback through comprehensive defect evaluation results.

[0090] This embodiment also provides a computer device, which is suitable for the case of a coating thickness detection method based on visual monitoring, including: a memory and a processor; the memory is used to store computer executable instructions, and the processor is used to execute computer executable instructions to implement the coating thickness detection method based on visual monitoring proposed in the above embodiment.

[0091] The computer device may be a terminal, and the computer device includes a processor, a memory, a communication interface, a display screen and an input device connected via a system bus. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The communication interface of the computer device is used to communicate with an external terminal in a wired or wireless manner, and the wireless manner can be achieved through WIFI, an operator network, NFC (near field communication) or other technologies. The display screen of the computer device may be a liquid crystal display screen or an electronic ink display screen, and the input device of the computer device may be a touch layer covering the display screen, or a key, trackball or touchpad provided on the housing of the computer device, or an external keyboard, touchpad or mouse, etc.

[0092] This embodiment also provides a storage medium on which a computer program is stored. When the program is executed by a processor, the coating thickness detection method based on visual monitoring as proposed in the above embodiment is implemented; the storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (Static Random Access Memory, referred to as SRAM), electrically erasable programmable read-only memory (Electrically Erasable Programmable Read-Only Memory, referred to as EEPROM), erasable programmable read-only memory (Erasable Programmable Read Only Memory, referred to as EPROM), programmable read-only memory (Programmable Red-Only Memory, referred to as PROM), read-only memory (Read-Only Memory, referred to as ROM), magnetic storage, flash memory, magnetic disk or optical disk.

[0093] In summary, the present invention significantly improves the prediction accuracy and adaptability by: real-time acquisition of coating surface images and preprocessing, combining geometric optics for preliminary thickness prediction, and then accurately predicting coating thickness through a multi-channel neural network. Furthermore, through the defect recognition and evaluation module, the type, shape, complexity and location of coating surface defects are comprehensively considered, and coating defects are accurately identified and quantitatively evaluated, thereby improving the accuracy and robustness of defect detection. Finally, real-time feedback is provided based on the comprehensive evaluation results, the operator is notified in a timely manner and relevant data is recorded to ensure that the coating quality is effectively controlled, optimize the quality management in the production process, and has broad application prospects.

[0094] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. A coating thickness detection method based on visual monitoring, characterized in that: include, Collect coating surface images in real time and pre-process the collected coating surface images; Preliminary prediction of coating thickness using geometric optics; Based on the preprocessed coating surface image and preliminary prediction value, the coating thickness is predicted through a multi-channel neural network; According to the predicted coating thickness, defects are identified and evaluated to generate the final coating thickness and defect comprehensive evaluation results; Provide real-time feedback through comprehensive defect assessment results.

2. The coating thickness detection method based on visual monitoring according to claim 1, characterized in that: The specific steps of real-time acquisition of coating surface images are as follows: The coating surface image is captured in real time using a high-resolution camera.

3. The coating thickness detection method based on visual monitoring according to claim 2, characterized in that: The collected coating surface image is preprocessed, and the specific steps are as follows: The collected images of the coating surface are subjected to denoising, image enhancement, illumination equalization, image segmentation, image alignment and geometric correction.

4. The coating thickness detection method based on visual monitoring according to claim 3, characterized in that: The specific steps of making a preliminary prediction of coating thickness by geometric optics are as follows: Select the reflection method as the calculation method, use the optical reflection coefficient of the coating and the incident light angle to calculate the thickness, receive the reflected light intensity on the coating surface, measure the ratio of the reflected light intensity to the incident light intensity, and obtain the reflectivity; Based on the principle of geometric optics, the coating thickness is preliminarily predicted as follows: Among them, H is the preliminary predicted coating thickness, λ is the wavelength of the light wave, and R is the reflectivity.

5. The coating thickness detection method based on visual monitoring according to claim 4, characterized in that: The coating thickness prediction is performed based on the preprocessed image and the preliminary prediction value through a multi-channel neural network. The specific steps are as follows: Based on the multi-channel neural network, the multi-channel neural network structure is divided into two independent channels. Channel 1 extracts the spatial features of the coating surface image through a deep convolutional neural network after preprocessing, and forms the feature vector of the image channel by splicing; Channel 2 inputs the preliminary prediction value into the multilayer perceptron, performs nonlinear mapping through the fully connected layer, and generates the feature vector of the preliminary prediction value; The feature vector of the image channel and the feature vector of the preliminary prediction value are weighted and combined through the weighted fusion layer to obtain the final fused feature vector. The final fused feature vector is used for the final coating thickness prediction through the fully connected layer to generate the predicted coating thickness.

6. The coating thickness detection method based on visual monitoring according to claim 5, characterized in that: The method of performing defect identification and evaluation based on the predicted coating thickness to generate the final coating thickness and defect evaluation results is as follows: Extract edge features of the preprocessed coating surface image, detect circular and linear defects on the coating surface, and perform segmentation and defect recognition on the coating surface in combination with morphological operations; The area and shape characteristics of the coating surface defect area are obtained through image analysis algorithms, and the complexity score of the coating surface defect is calculated based on the shape characteristics; The relative position impact of each defect area is defined by calculating the distance between the defect area and the edge of the coating. The coating surface defect severity score is calculated based on the comprehensive score of the defect area, the complexity score of the defect and the relative position impact score. The expression is: Among them, E is the severity score of coating surface defects, A is n is the area of ​​the nth coating surface defect area, A max is the maximum area of ​​the coating surface, C s is the complexity score of coating surface defects, s indicates that the complexity score of coating surface defects is related to the complexity of the geometric shape of the defects, C l is the relative position impact score, l means the relative position impact score is related to the position relationship of the defect relative to the coating edge; Based on the preliminary thickness predicted by geometric optics, the coating thickness predicted by the multi-channel neural network, and the coating surface defect severity score, the coating thickness is adjusted by a nonlinear function, and the final adjusted coating thickness is calculated, which is expressed as: Among them, H f is the final adjusted coating thickness, P is the coating thickness predicted by the multi-channel neural network, ΔH is the thickness error feedback, γ is the error correction coefficient, and α is the environmental correction coefficient; Based on the final adjusted coating thickness and the standard thickness of the coating, a coating quality assessment result is generated.

7. The coating thickness detection method based on visual monitoring according to claim 6, characterized in that: The above-mentioned comprehensive defect evaluation results are used to provide real-time feedback. The specific steps are as follows: Based on the comprehensive defect evaluation results, thresholds are set based on historical data. When the defect density and size are greater than the threshold, the coating quality is deemed unqualified. The operator is notified through a real-time alarm, and the coating thickness and defect score are recorded in real time.

8. A coating thickness detection system based on visual monitoring, based on the coating thickness detection method based on visual monitoring according to any one of claims 1 to 7, characterized in that: It includes image acquisition and preprocessing module, coating thickness preliminary prediction module, coating thickness prediction module, coating thickness and defect comprehensive evaluation module and real-time feedback module. An image acquisition and preprocessing module acquires coating surface images in real time and preprocesses the acquired coating surface images; The coating thickness preliminary prediction module makes a preliminary prediction of the coating thickness through geometric optics; The coating thickness prediction module predicts the coating thickness based on the preprocessed image and the preliminary prediction value through a multi-channel neural network; The coating thickness and defect comprehensive evaluation module identifies and evaluates defects based on the predicted coating thickness and generates the final coating thickness and defect comprehensive evaluation results; The real-time feedback module provides real-time feedback through comprehensive defect evaluation results.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the coating thickness detection method based on visual monitoring according to any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the coating thickness detection method based on visual monitoring according to any one of claims 1 to 7 are implemented.

Images