A method and system for identifying bubbles in inorganic decorative panel coatings based on images.

Abstract

This invention relates to the field of image processing technology, and more particularly to an image-based method and system for identifying bubbles in inorganic decorative panel coatings. The method includes the following steps: acquiring and preprocessing an image; then constructing a local texture anisotropy index and a local gradient energy index based on a structure tensor, synthesizing a defect structure saliency factor to distinguish between regular textures and irregular defects; subsequently calculating a texture consistency deviation index and converting it into a defect enhancement factor to weighted enhance the gradient image, suppressing background texture interference; finally, using adaptive threshold segmentation to extract bubble defects. This invention effectively suppresses interference from complex background textures by integrating local spatial structure information in multiple dimensions, significantly improving the accuracy and robustness of bubble defect detection in inorganic decorative panels.

Description

Technical Field

[0001] This invention relates to the field of image processing technology, and in particular to an image-based method and system for identifying bubbles in inorganic decorative panel coatings. Background Technology

[0002] Inorganic decorative panels, with their excellent properties such as fire resistance, moisture resistance, and wear resistance, are widely used in architectural decoration and home renovation. The smoothness and integrity of their surface coating directly determine the product's appearance quality and service life. In actual industrial production, fluctuations in process steps such as coating ratio accuracy, spray pressure stability, and curing temperature and time control can easily lead to bubble defects in the decorative panel surface coating. These bubbles not only affect the product's aesthetics but can also cause the coating to peel off, reducing the product's durability.

[0003] Currently, the detection of such defects mainly relies on machine vision technology, which uses industrial cameras to capture images of the decorative panel surface and edge detection and other techniques to detect bubble defects.

[0004] However, the principle of locating defects based on edge detection operators (such as the Canny operator) is to identify the edge contours of the defect area and the background area by calculating the gray-level gradient changes of image pixels, and then determine the location of the bubble defect. This approach has significant limitations when dealing with inorganic decorative panels with complex textures (such as imitation stone or wood grain). Since the complex texture of the decorative panel itself contains a large number of high-frequency details and randomly distributed stripes, its gray-level characteristics are very similar to those of the bubble defect, making it difficult for traditional algorithms to distinguish between the background texture and the bubble defect. This interference from the background texture makes edge detection operator-based methods prone to generating a large number of false detections, which cannot meet the needs of high-precision automated detection. Summary of the Invention

[0005] To address the technical problem that the Canny edge detection algorithm is prone to false detection due to the high similarity between the texture of inorganic decorative panels and the gradient features of bubble defects, thus failing to meet the requirements of high-precision automated detection, this invention provides an image-based method and system for identifying bubbles in inorganic decorative panel coatings.

[0006] In a first aspect, the present invention provides an image-based method for identifying bubbles in inorganic decorative panel coatings, employing the following technical solution:

[0007] A method for identifying air bubbles in an inorganic decorative panel coating based on an image, comprising the following steps:

[0008] Acquire images of the decorative panel surface;

[0009] A structure tensor matrix is ​​constructed based on the gradient information of each pixel in the surface image of the decorative panel; a local texture anisotropy index is obtained for each pixel based on the difference between the eigenvalues ​​of the structure tensor matrix of each pixel; a local gradient energy index is obtained for each pixel based on the gradient magnitude of the local region of the pixel; and a defect structure saliency factor for each pixel is obtained based on the local texture anisotropy index and the local gradient energy index.

[0010] Based on the difference in the saliency factor of the defect structure of a pixel and its local region pixels, a texture consistency deviation index is obtained for each pixel; a defect enhancement factor for each pixel is constructed based on the texture consistency deviation index; and the gradient magnitude is weighted using the defect enhancement factor of each pixel to obtain the enhanced gradient image of the decorative panel.

[0011] Based on the enhanced gradient image of the decorative panel, an adaptive threshold segmentation algorithm is used to extract the bubble region, thereby realizing the detection of bubble defects on the surface of inorganic decorative panels.

[0012] The innovation of this invention lies in obtaining the defect structure saliency factor of each pixel based on the local texture anisotropy index and the local gradient energy index, abandoning the traditional single-dimensional gradient detection mode. Based on the difference in eigenvalues ​​of the structure tensor matrix of the pixel, it can essentially distinguish the regular geometric attributes of the background texture from the irregular shape of the bubble defect. Then, based on the defect structure saliency factor, the texture consistency deviation index of each pixel is obtained to construct the defect enhancement factor of each pixel to correct the gradient magnitude, and the enhanced gradient image of the decorative panel is obtained for bubble recognition. The texture consistency deviation of each pixel is obtained by using the feature of the bubble disrupting the texture continuity. This method effectively solves the problem of the difficulty in distinguishing the nonlinear nodes of the background texture from the bubble defect, and achieves high-precision detection under high interference background.

[0013] Preferably, constructing the structure tensor matrix based on the gradient information of each pixel in the image includes:

[0014] Using the Sobel operator, the gradients of each pixel in the horizontal and vertical directions are obtained. Based on the gradients in the horizontal and vertical directions, the structure tensor matrix of each pixel is constructed, and the two eigenvalues ​​of the structure tensor matrix of each pixel are obtained.

[0015] Preferably, obtaining the local texture anisotropy index for each pixel includes:

[0016] ;

[0017] In the formula, The local texture anisotropy index represents the i-th pixel. The largest eigenvalue of the structure tensor matrix representing the i-th pixel; The smallest eigenvalue of the structure tensor matrix representing the i-th pixel; This represents a preset, small constant.

[0018] By quantifying the geometric structural properties of local regions using the magnitude relationship of eigenvalues, it is possible to accurately identify directional linear textures and isotropic bubble or dirt areas, providing key geometric feature basis for distinguishing background textures and potential defects.

[0019] Preferably, obtaining the local gradient energy index for each pixel includes:

[0020] Preset window length Build with each pixel as the center The window serves as a local region for each pixel;

[0021] , The local gradient energy index represents the i-th pixel. This represents the gradient magnitude of the j-th pixel within the local region of the i-th pixel. This represents the number of pixels in the local region of the i-th pixel. This represents the maximum gradient magnitude among all pixels in the image of the decorative panel surface.

[0022] It can distinguish bubble edge regions with high gradient amplitudes from dirty regions with low gradients, making up for the insufficiency of relying solely on anisotropy indicators to eliminate interference from dirty regions.

[0023] Preferably, obtaining the defect structure saliency factor for each pixel includes:

[0024] ;

[0025] In the formula, The defect structure saliency factor represents the i-th pixel. The local texture anisotropy index represents the i-th pixel. The local gradient energy index represents the i-th pixel. This represents the first weighted adjustment coefficient. This represents the second weighting adjustment coefficient.

[0026] Preliminary localization of pixels in the bubble region was achieved.

[0027] Preferably, obtaining the texture consistency deviation index for each pixel includes:

[0028] ;

[0029] In the formula, The texture consistency deviation index represents the i-th pixel. This represents the number of pixels in the local region of the i-th pixel. The defect structure saliency factor represents the i-th pixel. The defect structure saliency factor represents the j-th pixel in the local region of the i-th pixel; The mean of the defect structure saliency factor of all pixels in the local region of the i-th pixel; This represents the adjustment parameter.

[0030] Preferably, obtaining the defect enhancement factor for each pixel includes:

[0031] ;

[0032] In the formula, The defect enhancement factor represents the i-th pixel. The texture consistency deviation index represents the i-th pixel. Represents the logarithmic function; The sensitivity scaling factor represents the deviation value.

[0033] Preferably, acquiring the decorative panel enhancement gradient image includes:

[0034] The product of the gradient magnitude of each pixel and the defect enhancement factor of each pixel is used as the updated gradient magnitude of each pixel to obtain the decorative panel enhancement gradient image.

[0035] Background texture regions are suppressed due to the small enhancement factor, while bubble regions are significantly amplified due to the large original gradient and large enhancement factor. The resulting enhanced gradient image can accurately segment the bubble regions.

[0036] Preferably, the acquisition of the decorative panel surface image includes:

[0037] A high-resolution grayscale image of the inorganic decorative panel surface is acquired using a linear or area array industrial camera; the high-resolution grayscale image is then subjected to Gaussian smoothing filtering to remove random Gaussian noise generated during the acquisition process, thus obtaining the decorative panel surface image.

[0038] Secondly, the present invention provides an image-based inorganic decorative panel coating bubble recognition system, which adopts the following technical solution:

[0039] An image-based inorganic decorative panel coating bubble recognition system includes a processor and a memory, wherein the memory stores computer program instructions, and when the computer program instructions are executed by the processor, the aforementioned image-based inorganic decorative panel coating bubble recognition method is implemented.

[0040] By adopting the above technical solution, a computer program is generated from the above-mentioned image-based inorganic decorative panel coating bubble recognition method and stored in the memory so that it can be loaded and executed by the processor. In this way, a terminal device can be made based on the memory and the processor for convenient use.

[0041] This invention has the following technical effects: Based on local texture anisotropy index and local gradient energy index, this invention obtains the defect structure saliency factor, abandoning the traditional single-dimensional gradient detection mode. Based on the difference in eigenvalues ​​of the structure tensor matrix of pixels, it can fundamentally distinguish the regular geometric attributes of the background texture from the irregular shape of bubble defects. Furthermore, based on the defect structure saliency factor, it obtains the texture consistency deviation index of each pixel and constructs a defect enhancement factor for each pixel to correct the gradient magnitude, obtaining an enhanced gradient image of the decorative panel for bubble recognition. By utilizing the feature of bubbles disrupting texture continuity, it obtains the texture consistency deviation of each pixel. This method effectively solves the problem of distinguishing between nonlinear nodes of the background texture and bubble defects, achieving high-precision detection under high interference backgrounds. Attached Figure Description

[0042] Figure 1 This is a flowchart of an image-based method for identifying bubbles in inorganic decorative panel coatings according to an embodiment of the present invention.

[0043] Figure 2 It is a standard edge detection image;

[0044] Figure 3 It is a gradient image for enhancing the decorative panel;

[0045] Figure 4 This is the final bubble recognition result image. Detailed Implementation

[0046] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are some embodiments of the present invention, but not all embodiments.

[0047] This invention discloses an image-based method for identifying bubbles in inorganic decorative panel coatings, referring to... Figure 1 This includes steps S1-S4:

[0048] S1: Acquire an image of the decorative panel surface.

[0049] In this embodiment of the invention, a high-resolution grayscale image of the inorganic decorative panel surface is acquired by a linear or area array industrial camera.

[0050] Considering the influence of industrial lighting environment, the high-resolution grayscale image is subjected to Gaussian smoothing filtering to remove random Gaussian noise generated during the acquisition of the high-resolution grayscale image, so as to obtain the surface image of the decorative panel.

[0051] S2: Construct a structure tensor matrix based on the gradient information of each pixel in the image; obtain the local texture anisotropy index of each pixel based on the eigenvalues ​​of the structure tensor matrix of each pixel; obtain the local gradient energy index of each pixel based on the gradient magnitude of the local region of the pixel; obtain the defect structure saliency factor of each pixel based on the local texture anisotropy index and the local gradient energy index.

[0052] It should be noted that complex background textures in decorative panel surface images, such as the grain of wood grain and bubble defects, are highly confused in the gradient dimension. Traditional edge detection algorithms can only detect textures in the image, but cannot distinguish between the regular geometric structure of textures and the irregular shape of defects. Figure 2 As shown;

[0053] Gradient vectors can accurately characterize the direction and intensity of grayscale changes in pixels. Constructing a structure tensor matrix for each pixel based on gradient vectors can quantify the gradient distribution patterns within the pixel's neighborhood, thereby extracting the core geometric structural attributes of the local region. Specifically, the gradient vector distribution of background texture pixels exhibits a significant regular pattern, such as parallelism, and the corresponding local structure tensor matrix has a clear principal feature direction, with significant differences in eigenvalues. In contrast, the gradient vector distribution of bubble defect pixels is chaotic and disordered, lacking a unified principal feature direction, while the eigenvalue distribution of the local structure tensor matrix tends to be uniform. Therefore, this invention constructs a local texture anisotropy index for each pixel based on the eigenvalues ​​of its structure tensor matrix, enabling the distinction between regular textures and irregular defects.

[0054] In this embodiment of the invention, the Sobel operator is used to obtain the gradient of each pixel in the horizontal and vertical directions. Based on the gradients in the horizontal and vertical directions, the structure tensor matrix of each pixel is constructed, and two eigenvalues ​​of the structure tensor matrix of each pixel are obtained. It should be noted that the construction of the structure tensor matrix is ​​a well-known technique, and it will not be described in detail in this embodiment of the invention.

[0055] Obtain the local texture anisotropy index for each pixel:

[0056] ;

[0057] In the formula, The local texture anisotropy index represents the i-th pixel. The largest eigenvalue of the structure tensor matrix representing the i-th pixel; Let represent the minimum eigenvalue of the structure tensor matrix of the i-th pixel. It should be noted that... ; Representing a preset infinitesimal constant, in this embodiment of the invention, the preset... To prevent the denominator from being zero;

[0058] The value of is much greater than When this occurs, it indicates that the i-th pixel has obvious linear texture features (such as wood grain patterns). Approaching 1; The value of and When they are close, it indicates that the i-th pixel is located in a flat area, or may be a bubble defect area or a dirty area. Approaching 0;

[0059] It should be noted that inorganic dust (such as cement-based powder and stone powder) is generated during the forming, polishing, and cutting processes of decorative panels. These fine dust particles may adhere evenly to the panel surface, forming dirty areas. Since the gradient vector distribution of dirty areas is chaotic and disordered, the local texture anisotropy index cannot distinguish between pixels in bubble defect areas and dirty pixels. Because bubble defect areas are raised or recessed, they appear to have reflections and shadows under lighting conditions, resulting in extremely sharp edges and extremely high gradient energy. However, the edges of dirty areas are blurred and the gradient energy is low. Therefore, this invention constructs a local gradient energy index based on the gradient amplitude of local areas of pixels to distinguish between the two.

[0060] In this embodiment of the invention, a preset window length is used. Build with each pixel as the center The window serves as a local region for each pixel;

[0061] Obtain the local gradient energy index for each pixel:

[0062] ;

[0063] In the formula, The local gradient energy index represents the i-th pixel. This represents the gradient magnitude of the j-th pixel within the local region of the i-th pixel. This represents the number of pixels in the local region of the i-th pixel. This represents the maximum gradient magnitude among all pixels in the decorative panel surface image, used for normalization.

[0064] When the gradient magnitude is large in a local region of a pixel, The larger the value, the greater the saliency of the texture in the local area, indicating that the pixel is more likely to belong to the bubble edge and background texture; conversely, when the gradient magnitude in the local area of ​​the pixel is small, it indicates that the pixel is more likely to belong to the flat background and dirty area.

[0065] It should be noted that this invention combines local texture anisotropy index and local gradient energy index to obtain the defect structure significance factor. Only pixels with smaller local texture anisotropy index and larger local gradient energy index are more likely to belong to bubble defect pixels.

[0066] In this embodiment of the invention, the defect structure saliency factor of each pixel is obtained:

[0067] ;

[0068] In the formula, The defect structure saliency factor represents the i-th pixel. The local texture anisotropy index represents the i-th pixel. The local gradient energy index represents the i-th pixel. This represents the first weighted adjustment coefficient. This represents the second weighting adjustment coefficient; in this embodiment of the invention, a preset... , In other embodiments, implementers may pre-determine specific implementation methods. as well as The value;

[0069] The round shape of the bubbles leads to The value approaches 0, at which point... The larger the value, the more reflective the bubble edges become. The higher the value, the greater the value of the defect structure salience factor of the pixels in the bubble region, due to the superposition of the two products.

[0070] Texture area pixels cause The value approaches 1, causing... The value of approaches 0, therefore the number of pixels in the texture region... The smaller the value;

[0071] Flat or dirty areas cause The value approaches 0, at which point... The larger the value, the smaller the local gradient energy index of pixels in flat or dirty regions. Therefore, pixels in flat or dirty regions... The smaller the value.

[0072] S3: Based on the defect structure saliency factor, obtain the texture consistency deviation index of each pixel; construct the defect enhancement factor of each pixel according to the texture consistency deviation index; use the defect enhancement factor of each pixel to weight the gradient magnitude and generate the decorative panel enhancement gradient image.

[0073] It should be noted that for non-linear nodes in the background texture (such as the bifurcations of wood grain), the gradient directions within them are randomly distributed due to the randomness of the texture structure itself, without a unified main direction, and the gradient amplitude is relatively large. Therefore, the larger the value of the defect structure saliency factor, the more likely it is to be identified as a non-linear region. Relying solely on the defect structure saliency factor cannot identify non-linear nodes and bubble defects in the background texture. It is known that non-linear nodes in the background texture are natural components of the background texture, and they are spatially continuous with the surrounding texture (for example, bifurcations connect the wood grain texture after the bifurcation and the surrounding small bifurcations). The gradient directions of the wood grain texture after the bifurcation and the surrounding small bifurcations are also randomly distributed. Therefore, the difference in the defect structure saliency factor between the pixel at the bifurcation node and the pixel in its local area is relatively small. On the other hand, bubble defects are raised areas on inorganic decorative panels, which will break the continuity of the surrounding texture (the local pixels of bubble defects are regular background textures or flat areas). The difference in the defect structure saliency factor between the bubble defect pixel and the pixel in its local area is larger.

[0074] Therefore, this invention obtains a texture consistency deviation index based on the difference between the defect structure significance factor of a pixel and the pixels in its local region. The larger the value, the more likely the pixel is to be a bubble defect pixel.

[0075] In this embodiment of the invention, the texture consistency deviation index of each pixel is obtained:

[0076] ;

[0077] In the formula, The texture consistency deviation index represents the i-th pixel. This represents the number of pixels in the local region of the i-th pixel. The defect structure saliency factor represents the i-th pixel. The defect structure saliency factor represents the j-th pixel in the local region of the i-th pixel; The mean of the defect structure saliency factor of all pixels in the local region of the i-th pixel; Represents the adjustable parameter, preset. In other embodiments, implementers may pre-set according to specific implementation conditions. The value; The larger the value, the higher the texture consistency deviation index of the pixel, and the more likely the pixel is to be a defective pixel.

[0078] The degree of abrupt change is characterized by calculating the difference in the saliency factor of the defect structure between a pixel and its local region pixels. For nonlinear nodes in normal background textures, the saliency factor of the defect structure is similar in a local range (i.e., the texture has continuity), and the difference term is small, leading to... The value is relatively low; however, bubble defects, as a sudden morphological defect, disrupt the continuity of the surrounding texture, causing significant differences in the defect structure significance factor within a local range, thus... A sharp increase.

[0079] It should be noted that in order to accurately identify the bubble region, the texture consistency deviation index mentioned above needs to be mapped to a defect enhancement factor, which is used to amplify the gradient value of the bubble region in the subsequent process, so that the bubble region can be accurately segmented.

[0080] In this embodiment of the invention, the defect enhancement factor for each pixel is obtained:

[0081] ;

[0082] In the formula, The defect enhancement factor represents the i-th pixel. The texture consistency deviation index represents the i-th pixel. Represents the logarithmic function; The sensitivity scaling factor representing the deviation value is preset in this embodiment of the invention. In other embodiments, implementers may pre-determine specific implementation methods. The value, The smaller the value, the more sensitive it is to larger values ​​of texture consistency deviation, indicating very significant texture corruption. The larger the value, the larger the output. ;

[0083] The larger the value, the more likely the pixel is to be a bubble defect, and the larger the defect enhancement factor of the pixel is; this formula uses a logarithmic function to map the texture consistency deviation index, when the independent variable... When it increases, The value increases, but the growth rate gradually slows down, thereby providing a significant gain to bubble areas with large texture consistency deviations, while smoothing extreme noise values.

[0084] The product of the gradient magnitude of each pixel and the defect enhancement factor of each pixel is used as the updated gradient magnitude of each pixel to obtain the decorative panel enhancement gradient image.

[0085] Thus, through multiplicative weighting, the background texture region is suppressed due to its smaller defect enhancement factor, while the bubble region is doubly enhanced due to its larger defect enhancement factor and larger original gradient. Figure 3 As shown.

[0086] S4: Based on the enhanced gradient image of the decorative panel, an adaptive threshold segmentation algorithm is used to extract the bubble region, thereby realizing the detection of bubble defects on the surface of inorganic decorative panels.

[0087] In this embodiment of the invention, based on the enhanced gradient image of the decorative panel, the segmentation threshold is adaptively calculated using the maximum inter-class variance method. The enhanced gradient image of the decorative panel is then binarized to extract bubble defects, such as... Figure 4 As shown.

[0088] This invention also discloses an image-based inorganic decorative panel coating bubble recognition system, including a processor and a memory. The memory stores computer program instructions, which, when executed by the processor, implement an image-based inorganic decorative panel coating bubble recognition method provided by this invention.

[0089] The system also includes other components well-known to those skilled in the art, such as communication buses and communication interfaces, the setup and functions of which are known in the art and will not be described in detail here. In this invention, the aforementioned memory can be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, apparatus, or device.

[0090] The above are all preferred embodiments of the present invention and are not intended to limit the scope of protection of the present invention. Therefore, all equivalent changes made in accordance with the structure, shape and principle of the present invention should be covered within the scope of protection of the present invention.

Claims

1. An image-based inorganic finish panel coating bubble identification method, characterized by, include: Acquire images of the decorative panel surface; A structural tensor matrix is ​​constructed based on the gradient information of each pixel in the surface image of the decorative panel. Based on the differences between the eigenvalues ​​of the structure tensor matrix of each pixel, the local texture anisotropy index of each pixel is obtained; based on the gradient magnitude of the local region of each pixel, the local gradient energy index of each pixel is obtained, including: preset window length. Build with each pixel as the center The window serves as a local region for each pixel; , Let be the local gradient energy index of the i-th pixel. Let be the gradient magnitude of the j-th pixel in the local region of the i-th pixel. Let be the number of pixels in the local region of the i-th pixel. This represents the maximum gradient magnitude among all pixels in the image of the decorative panel surface. Based on local texture anisotropy indices and local gradient energy indices, the defect structure saliency factor of each pixel is obtained, including: , Let be the defect structure saliency factor of the i-th pixel. Let be the local texture anisotropy index of the i-th pixel. This is the first weighted adjustment coefficient. This is the second weighting adjustment coefficient; Based on the difference in the saliency factor of the defect structure between a pixel and its local region pixels, the texture consistency deviation index of each pixel is obtained, including: , Let be the texture consistency deviation index for the i-th pixel. Let be the number of pixels in the local region of the i-th pixel. Let be the defect structure saliency factor of the j-th pixel in the local region of the i-th pixel. Let be the mean of the defect structure saliency factor of all pixels in the local region of the i-th pixel. To adjust the parameters; According to the texture consistency deviation index, a defect enhancement factor of each pixel point is constructed, including: , is a defect enhancement factor of the i-th pixel point, is a logarithmic function, is a sensitivity scaling coefficient of the deviation value; The gradient magnitude is weighted using the defect enhancement factor of each pixel to obtain the enhanced gradient image of the decorative panel; Based on the enhanced gradient image of the decorative panel, an adaptive threshold segmentation algorithm is used to extract the bubble region, thereby realizing the detection of bubble defects on the surface of inorganic decorative panels.

2. The image-based inorganic veneer panel coating bubble identification method of claim 1, wherein, The construction of the structure tensor matrix based on the gradient information of each pixel in the image includes: Using the Sobel operator, the gradients of each pixel in the horizontal and vertical directions are obtained. Based on the gradients in the horizontal and vertical directions, the structure tensor matrix of each pixel is constructed, and the two eigenvalues ​​of the structure tensor matrix of each pixel are obtained.

3. The image-based inorganic veneer panel coating bubble identification method of claim 1, wherein, The process of obtaining the local texture anisotropy index for each pixel includes: ； In the formula, representing a local texture anisotropy index of the i-th pixel point; representing a maximum eigenvalue of a structure tensor matrix of the i-th pixel point; representing a minimum eigenvalue of the structure tensor matrix of the i-th pixel point; representing a preset infinitesimal constant.

4. The image-based inorganic veneer panel coating bubble identification method of claim 1, wherein, The acquisition of the decorative panel enhancement gradient image includes: The product of the gradient magnitude of each pixel and the defect enhancement factor of each pixel is used as the updated gradient magnitude of each pixel to obtain the decorative panel enhancement gradient image.

5. The image-based inorganic veneer panel coating bubble identification method of claim 1, wherein, The acquisition of images of the decorative panel surface includes: A high-resolution grayscale image of the inorganic decorative panel surface is acquired using a linear or area array industrial camera; the high-resolution grayscale image is then subjected to Gaussian smoothing filtering to remove random Gaussian noise generated during the acquisition process, thus obtaining the decorative panel surface image.

6. An image-based inorganic veneer panel coating bubble identification system, comprising: include: A processor and a memory, the memory storing computer program instructions that, when executed by the processor, implement an image-based method for identifying bubbles in inorganic decorative panel coatings according to any one of claims 1-5.

Images