A fire mask production process intelligent monitoring method and system

Abstract

The present invention relates to the field of image processing technology. Specifically, it relates to a method and system for intelligently monitoring the production process of a fire mask. The method comprises: collecting an image of a fire mask and graying it to obtain a grayscale image of the fire mask; establishing a window of a preset size with each pixel of the grayscale image as the center; calculating the noise performance degree of the central pixel of each window; filtering each central pixel using a bilateral filtering algorithm, including: correcting the comprehensive weight corresponding to the central pixel according to the noise performance degree to obtain a comprehensive weight correction value, wherein the comprehensive weight correction value is negatively correlated with the noise performance degree; and inputting the grayscale image of the fire mask after filtering into a preset defect recognition model. The method and system for intelligently monitoring the production process of a fire mask of the present invention can improve the accuracy of defect recognition of fire masks.

Description

Technical Field

[0001] The present invention relates to the field of image processing technology, and more specifically, to an intelligent monitoring method and system for a fire mask production process. Background Art

[0002] In a fire or toxic and harmful environment, a large amount of smoke and toxic gases will be produced. Fire masks can provide people with reliable protection, prevent poisoning caused by smoke and high temperature burns, and reduce casualties caused by fire. Fire masks can effectively protect people's respiratory organs, head and face, and fire masks can provide wearers with a clear field of vision, allowing people to evacuate safely from fires or toxic and harmful environments. However, if there are defects in the fire mask, it cannot effectively protect the wearer in a fire or toxic and harmful environment, resulting in a threat to the health and safety of the wearer. Therefore, it is very important to monitor the quality of the fire mask in real time during the production process of the fire mask.

[0003] In the prior art, the method for real-time monitoring of the quality of products produced on an assembly line is to collect image data of the products and input it into a defect detection model to identify defects. However, if the collected image data contains noise, the accuracy of the defect identification result will be poor. For example, the paint defect detection method disclosed in the Chinese patent application document with application publication number CN117409010A collects paint images of products and inputs them into a defect detection model for defect detection, thereby identifying paint defects on the surface of the product. However, since the collected paint images are not denoised, when there are noise points in the collected images, the accuracy of the defect identification result will be poor. Another example is the workpiece surface defect detection method disclosed in the Chinese patent with authorization announcement number CN107966447B. It first collects images of each surface of the workpiece, and inputs the collected images of the workpiece to be tested into a trained convolutional neural network to obtain a defect area marking image. However, when there are noise points in the collected images, the accuracy of its defect identification result will also be poor.

[0004] In order to filter and denoise the collected images of fire masks and make the collected images clearer, the bilateral filtering method can be used to filter the collected images. However, since there is texture on the fire mask, the texture has various surfaces and surface edges, and there is a certain change in grayscale on the texture surface or on the surface edge, when there is noise on the texture or on the edge of the texture, when using traditional bilateral filtering for denoising, due to the grayscale change of the texture edge, for traditional bilateral filtering, only considering the distance and grayscale difference is not enough to denoise the pixels on the texture surface with grayscale changes or the pixels on the edge of the surface, so that the traditional bilateral filtering denoising effect is not good, which in turn leads to poor defect detection effect on the produced fire masks. Summary of the invention

[0005] In order to solve the technical problem of low accuracy when detecting defects of fire masks using methods in the prior art, the present invention provides solutions in the following aspects.

[0006] In a first aspect, the present invention provides a method for intelligently monitoring a fire mask production process, comprising the following steps:

[0007] Collect the image of the fire mask and grayscale it to obtain a grayscale image of the fire mask;

[0008] Establishing a window of a preset size with each pixel of the grayscale image as the center;

[0009] Calculate the noise level of the central pixel of each window. The calculation expression is: ;

[0010] In the formula, Indicates the number of non-center pixels in the window. Indicates the grayscale deviation of the central pixel in the i-th window. It indicates the grayscale deviation degree of the central pixel in the i-th window from the adjacent pixels in its gradient direction. Indicates the grayscale deviation of the jth non-center pixel in the i-th window. Indicates the grayscale deviation degree of the jth non-central pixel point in the i-th window from the adjacent pixel points in the gradient direction; the grayscale deviation degree is used to characterize the grayscale difference between a certain pixel point and other pixel points in the window; Indicates the noise performance of the central pixel of the i-th window;

[0011] Filtering each central pixel point using a bilateral filtering algorithm, including: correcting the comprehensive weight corresponding to the central pixel point according to the noise performance degree to obtain a comprehensive weight correction value, wherein the comprehensive weight correction value is negatively correlated with the noise performance degree;

[0012] The filtered fire mask grayscale image is input into the preset defect recognition model.

[0013] The beneficial effects are as follows: the intelligent monitoring method for the production process of fire masks of the present invention first uses a bilateral filtering algorithm to denoise the image before inputting the collected fire mask image into the defect recognition model. During denoising, the noise performance degree of the central pixel point is first calculated, and then the convolution kernel calculation of the bilateral filter is weighted according to the noise performance degree, and the weighted comprehensive weight is inversely proportional to the noise performance degree. The central pixel point is filtered according to the weighted comprehensive weight, thereby avoiding the situation in which the edge details in the fire mask production process are not obvious due to the blur caused by the motion of the collected image, so as to ensure that the noise influence of the image is accurately eliminated and the texture edge details are retained, thereby improving the accuracy of fire mask defect recognition, and further achieving more accurate monitoring of the fire mask production process.

[0014] Optionally, the calculation expression of the comprehensive weight correction value is:

[0015] ;

[0016] In the formula, represents the comprehensive weight correction value, represents the horizontal coordinate of the center pixel, j represents the vertical coordinate of the center pixel, Indicates the horizontal coordinate of any pixel point in the corresponding window. Indicates the ordinate of any pixel point in the corresponding window. represents the natural exponential function, Represents the gray value of the center pixel, The horizontal axis is , the vertical axis is The gray value of the pixel, The coordinates are ( k , l ) is the normalized noise level of the pixel point, represents the spatial weight parameter, Represents the intensity weight parameter.

[0017] Optionally, for a central pixel in a window, the calculation expression for the grayscale deviation is:

[0018] ;

[0019] In the formula, It indicates the grayscale difference consistency of the jth non-central pixel in the i-th window before the central pixel is removed. It indicates the grayscale difference consistency of the jth non-central pixel in the i-th window after the central pixel is removed. It represents the change in the grayscale difference consistency of the jth non-central pixel after the central pixel in the window is removed. represents the variance function, represents the variance of the change in the grayscale difference consistency, j=1,...,n-1, and the grayscale difference consistency is used to characterize the consistency between the grayscale difference of the pixel and the remaining pixels in the window, It represents the grayscale deviation of the central pixel in the i-th window, and n represents the number of pixels in the window.

[0020] The beneficial effect is that the calculation expression of the grayscale deviation degree of the present invention not only takes into account the change in grayscale difference consistency corresponding to the non-central pixel points in the window after the central pixel point is removed when calculating the grayscale deviation degree of the central pixel point, but also takes into account the different degrees of influence of the central pixel point on the change in grayscale difference consistency of other pixel points in the window, thereby making the calculation result of the grayscale deviation degree more accurate.

[0021] Optionally, for the central pixel in the i-th window, the calculation expression of its grayscale deviation degree is:

[0022] ;

[0023] In the formula, Indicates the chaos degree of the pixels in the i-th window, It indicates the grayscale difference consistency of the jth non-central pixel in the i-th window before the central pixel is removed. It indicates the grayscale difference consistency of the jth non-central pixel in the i-th window after the central pixel is removed. It represents the change in the grayscale difference consistency of the jth non-central pixel after the central pixel in the window is removed. represents the variance function, The variance of the change in the grayscale difference consistency is represented, j=1, ..., n-1, and the chaos degree is used to characterize the grayscale value difference between different pixels in the window. It represents the grayscale deviation of the central pixel in the i-th window, and n represents the number of pixels in the window.

[0024] The beneficial effect is that the greater the difference in grayscale values ​​between different pixels in the window, the greater the possibility that the central pixel is a texture pixel or noise. When calculating the grayscale deviation of the central pixel, not only the change in grayscale difference consistency corresponding to the non-central pixels in the window after the central pixel is removed, but also the different degrees of influence of the central pixel on the change in grayscale difference consistency of other pixels in the window are taken into account, but also the chaos of the pixels in the window is further taken into account, thereby further improving the accuracy of the grayscale deviation calculation result of the central pixel.

[0025] Optionally, for a pixel point in a window, the method for calculating the grayscale difference consistency includes: respectively calculating the difference between the grayscale value of the pixel point and the grayscale values ​​of all other pixel points in the window, and then calculating the variance of the difference, and taking the inverse of the variance of the difference as the grayscale difference consistency of the pixel point.

[0026] Its beneficial effect is: by calculating the variance of the difference between the grayscale value of the pixel point and the grayscale values ​​of all other pixel points in the window, the larger the variance, the greater the fluctuation of the difference, and the more inconsistent the grayscale difference between the grayscale value of the pixel point and the grayscale values ​​of all other pixel points in the window. The grayscale difference consistency calculation method of the present invention can more accurately calculate the grayscale difference consistency of the pixel points.

[0027] Optionally, the calculation expression of the disorder degree of pixels in a certain window is:

[0028] ;

[0029] In the formula, Indicates the chaos degree of the pixels in the i-th window, Indicates the number of pixels in the window. Represents the gray value of the central pixel in the i-th window, Represents the gray value of the j-th pixel in the i-th window.

[0030] Its beneficial effect is: the grayscale value difference between the central pixel and each pixel in the window is used to measure the chaos of the pixels in the window. The greater the grayscale value difference between the central pixel and each pixel in the window, the greater the chaos of the pixels in the window. When calculating the chaos of the pixels in the window, the average of all calculated grayscale value differences is taken as the chaos of the pixels in the window, so that the calculated chaos is more accurate and objective.

[0031] Optionally, the calculation expression of the disorder degree of pixels in a certain window is:

[0032] ;

[0033] In the formula, Represents the variance of the grayscale value of the pixel in the i-th window, Indicates the chaos degree of the pixels in the i-th window, Indicates the number of pixels in the window. Represents the gray value of the central pixel in the i-th window, Represents the gray value of the j-th pixel in the i-th window.

[0034] The beneficial effect is that the larger the variance corresponding to the grayscale of all pixels in the window, the greater the possibility of texture or noise influence in the window, and the greater the chaos of the pixels in the window. When calculating the chaos of the pixels in the window, not only the difference in grayscale values ​​between the central pixel and each pixel in the window is taken into account, but also the variance corresponding to the grayscale of all pixels in the window is further taken into account, so that the calculation result of the chaos of the pixels in the window is more accurate.

[0035] Optionally, the size of the window is 5 5.

[0036] In a second aspect, the present invention provides an intelligent monitoring system for a fire mask production process, comprising a processor and a memory, wherein the memory stores computer program instructions, and when the computer program instructions are executed by the processor, the intelligent monitoring method for a fire mask production process of the present invention is implemented.

[0037] In summary, the technical effect of the present invention is: the intelligent monitoring method of the fire mask production process of the present invention can greatly improve the accuracy of fire mask defect identification, thereby realizing more accurate monitoring of the fire mask production process. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] By reading the following detailed description with reference to the accompanying drawings, the above and other objects, features and advantages of the exemplary embodiments of the present invention will become readily understood. In the accompanying drawings, several embodiments of the present invention are shown in an exemplary and non-limiting manner, and the same or corresponding reference numerals represent the same or corresponding parts, wherein:

[0039] Figure 1 is a flow chart schematically showing a method for intelligent monitoring of a fire mask production process according to an embodiment of the present invention;

[0040] Figure 2 is a schematic diagram schematically showing the fire mask defect recognition result of an embodiment of the present invention;

[0041] Figure 3 It is a schematic diagram showing the structure of the intelligent monitoring system for the production process of fire masks according to an embodiment of the present invention. DETAILED DESCRIPTION

[0042] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative work are within the scope of protection of the present invention.

[0043] The specific embodiments of the present invention are described in detail below with reference to the accompanying drawings.

[0044] Implementation example of intelligent monitoring method for fire mask production process:

[0045] like Figure 1 As shown, the intelligent monitoring method for the production process of fire masks of the present invention comprises:

[0046] S101, collecting an image of a fire mask and graying it to obtain a grayscale image of the fire mask.

[0047] S102, establishing a window of a preset size with each pixel of the grayscale image as the center;

[0048] S103, calculating the noise performance degree of the central pixel of each window, the calculation expression is:

[0049]

[0050] In the formula, Indicates the number of non-center pixels in the window. Indicates the grayscale deviation of the central pixel in the i-th window. It indicates the grayscale deviation degree of the central pixel in the i-th window from the adjacent pixels in its gradient direction. Indicates the grayscale deviation of the jth non-center pixel in the i-th window. Indicates the grayscale deviation degree of the jth non-central pixel point in the i-th window from the adjacent pixel points in the gradient direction; the grayscale deviation degree is used to characterize the grayscale difference between a certain pixel point and other pixel points in the window; Indicates the approximation between the central pixel of the i-th window and the pixel in the gradient direction; Indicates the similarity between the jth non-central pixel of the i-th window and its adjacent pixels in the gradient direction of the central pixel; Indicates the degree of noise performance of the central pixel of the i-th window.

[0051] S104, using a bilateral filtering algorithm to filter each central pixel point, including: correcting the comprehensive weight corresponding to the central pixel point according to the noise performance level to obtain a comprehensive weight correction value, wherein the comprehensive weight correction value is negatively correlated with the noise performance level.

[0052] When using the traditional bilateral filtering algorithm to filter an image, it is necessary to calculate the spatial domain weight and the pixel value domain weight, and multiply the two to obtain a comprehensive weight; then, the pixel values ​​in the corresponding window are weighted averaged according to the comprehensive weight to obtain the pixel value of the center pixel after filtering.

[0053] The greater the noise performance of the central pixel, the greater the possibility that the central pixel is noise. This embodiment corrects the comprehensive weight of the central pixel according to the noise performance, and the greater the noise performance, the smaller the comprehensive weight; so that the pixel blurred by motion has more weight than the noise, and the closer to the actual pixel, the smaller the noise level value, and the higher the weight in the filter, so as to prevent the loss of filtering edge details due to edge blur.

[0054] S105, inputting the filtered fire mask grayscale image into a preset defect recognition model.

[0055] The defect recognition model can be a neural network model or other suitable models.

[0056] The defect recognition result obtained by inputting the denoised fire mask grayscale image into the defect recognition model is as follows: Figure 2 As shown, defect A is a particle defect and defect B is a recessed defect.

[0057] The intelligent monitoring method for the production process of fire masks of the present invention calculates the noise performance degree of the central pixel point of the window firstly calculates the similarity between the central pixel point and the pixel points adjacent to it in the gradient direction, and the similarity between the non-central pixel point and the pixel points adjacent to it in the gradient direction of the central pixel point, and calculates the noise performance degree of the central pixel point according to the difference between the two similarities, so that the calculation of the noise performance degree is more accurate, and after calculating the noise performance degree, the convolution kernel calculation of the bilateral filter is weighted according to the noise performance degree, and the weighted comprehensive weight is inversely proportional to the noise performance degree, and the central pixel point is filtered according to the weighted comprehensive weight, so as to avoid the situation in which the edge details in the production process of the fire mask are not obvious due to the blur caused by the motion of the collected image, so as to ensure that the noise influence of the image is accurately eliminated and the texture edge details are retained; after the mask image is denoised, it is input into the defect recognition model to improve the accuracy of fire mask defect recognition, thereby realizing more accurate monitoring of the fire mask production process.

[0058] In one embodiment, the calculation expression of the comprehensive weight correction value is:

[0059]

[0060] In the formula, represents the comprehensive weight correction value, represents the horizontal coordinate of the center pixel, j represents the vertical coordinate of the center pixel, Indicates the horizontal coordinate of any pixel point in the corresponding window. Indicates the ordinate of any pixel point in the corresponding window. represents the natural exponential function, Represents the gray value of the center pixel, The horizontal axis is , the vertical axis is The gray value of the pixel, The coordinates are The normalized noise level of the pixel point, represents the spatial weight parameter, Represents the intensity weight parameter.

[0061] In one embodiment, for a central pixel in a certain window, the calculation expression of its grayscale deviation degree is:

[0062]

[0063] In the formula, It indicates the grayscale difference consistency of the jth non-central pixel in the i-th window before the central pixel is removed. It indicates the grayscale difference consistency of the jth non-central pixel in the i-th window after the central pixel is removed. It represents the change in the grayscale difference consistency of the jth non-central pixel after the central pixel in the window is removed. represents the variance function, represents the variance of the change in the grayscale difference consistency, j=1,...,n-1, and the grayscale difference consistency is used to characterize the consistency between the grayscale difference of the pixel and the remaining pixels in the window, Indicates the grayscale deviation degree of the central pixel in the i-th window, and n indicates the number of pixels in the window. For example: assuming that the remaining pixels in the window include pixel A, pixel B and pixel C, the grayscale difference between the pixel and pixel A is a, the grayscale difference between the pixel and pixel B is b, and the grayscale difference between the pixel and pixel C is c, then the consistency between the grayscale differences between the pixel and the remaining pixels in the window refers to the consistency between a, b and c. In this embodiment, the consistency can be determined by the degree of proximity between the three. The closer the values ​​of the three are, the higher the consistency.

[0064] If the change in the grayscale difference consistency of the jth pixel in the ith window is large after the central pixel is removed, it means that the central pixel has a greater influence on the grayscale difference consistency of the jth pixel. The greater the grayscale deviation between the central pixel and the remaining pixels in the window, the greater the grayscale deviation of the central pixel. By calculating the change in the grayscale difference consistency corresponding to each non-central pixel in the window after the central pixel is removed and summing them up, all non-central pixels in the window are comprehensively considered when measuring the grayscale deviation of the central pixel, avoiding the problem of generalizing from a single case, and making the grayscale deviation evaluation result of the central pixel more accurate. It reflects the different degrees of influence of the central pixel on the grayscale difference consistency change of other pixels in the window. The larger the value, the greater the deviation between the pixel and other pixels in the window. The variance is used to determine whether the differences between the pixels in the window and the central pixel are similar. When the differences are uneven, it means that the deviation of other pixels in the window is large, and the possibility that the pixel is a texture edge pixel or a noise pixel is greater. The calculation expression of the grayscale deviation degree of this embodiment not only takes into account the change in grayscale difference consistency corresponding to the non-central pixels in the window after the central pixel is removed when calculating the grayscale deviation degree of the central pixel, but also takes into account the different degrees of influence of the central pixel on the grayscale difference consistency change of other pixels in the window, thereby making the calculation result of the grayscale deviation degree more accurate.

[0065] In the above embodiments, when calculating the grayscale deviation degree of the central pixel, the sum and variance of the changes in the grayscale difference consistency of the pixels in the window after the central pixel is removed are used; in one embodiment, for the central pixel in the i-th window, the calculation expression for the grayscale deviation degree is:

[0066]

[0067] In the formula, Indicates the chaos degree of the pixels in the i-th window, It indicates the grayscale difference consistency of the jth non-central pixel in the i-th window before the central pixel is removed. It indicates the grayscale difference consistency of the jth non-central pixel in the i-th window after the central pixel is removed. It represents the change in the grayscale difference consistency of the jth non-central pixel after the central pixel in the window is removed. represents the variance function, The variance of the change in the grayscale difference consistency is represented, j=1, ..., n-1, and the chaos degree is used to characterize the grayscale value difference between different pixels in the window. It represents the grayscale deviation of the central pixel in the i-th window, and n represents the number of pixels in the window.

[0068] The greater the difference in grayscale values ​​between different pixels in the window, the greater the possibility that the central pixel is a texture pixel or noise. In this embodiment, when calculating the grayscale deviation of the central pixel, not only the change in grayscale difference consistency corresponding to the non-central pixels in the window after the central pixel is removed, but also the different degrees of influence of the central pixel on the change in grayscale difference consistency of other pixels in the window are taken into account, but also the degree of disorder of the pixels in the window is further taken into account, thereby further improving the accuracy of the grayscale deviation calculation result of the central pixel.

[0069] In one embodiment, for a pixel point in a window, a method for calculating its grayscale difference consistency includes: respectively calculating the difference between the grayscale value of the pixel point and the grayscale values ​​of all other pixel points in the window, and then calculating the variance of the difference, and taking the inverse of the variance of the difference as the grayscale difference consistency of the pixel point.

[0070] The grayscale difference consistency calculation method of the present embodiment calculates the variance of the difference between the grayscale value of the pixel point and the grayscale values ​​of all other pixel points in the window. The larger the variance, the greater the fluctuation of the difference, and the more inconsistent the grayscale value of the pixel point and the grayscale difference between all other pixel points in the window. The grayscale difference consistency calculation method of the present embodiment can more accurately calculate the grayscale difference consistency of the pixel point.

[0071] In one embodiment, the calculation expression of the disorder degree of pixels in a certain window is:

[0072]

[0073] In the formula, Indicates the chaos degree of the pixels in the i-th window, Indicates the number of pixels in the window. Represents the gray value of the central pixel in the i-th window, Represents the gray value of the j-th pixel in the i-th window.

[0074] In this embodiment, the grayscale value difference between the central pixel point and each pixel point in the window is used to measure the chaos of the pixels in the window. The greater the grayscale value difference between the central pixel point and each pixel point in the window, the greater the chaos of the pixels in the window. When calculating the chaos of the pixels in the window, the average of all calculated grayscale value differences is taken as the chaos of the pixels in the window, so that the calculated chaos is more accurate and objective.

[0075] In one embodiment, the calculation expression of the disorder degree of pixels in a certain window is:

[0076]

[0077] In the formula, Represents the variance of the grayscale value of the pixel in the i-th window, Indicates the chaos degree of the pixels in the i-th window, Indicates the number of pixels in the window. Represents the gray value of the central pixel in the i-th window, Represents the gray value of the j-th pixel in the i-th window.

[0078] The larger the variance corresponding to the grayscale of all pixels in the window, the greater the possibility of texture or noise influence in the window, and the greater the chaos of the pixels in the window. In this embodiment, when calculating the chaos of the pixels in the window, not only the difference in grayscale value between the central pixel and each pixel in the window is taken into account, but also the variance corresponding to the grayscale of all pixels in the window is further taken into account, so that the calculation result of the chaos of the pixels in the window is more accurate.

[0079] In one embodiment, the size of the window is 5 5.

[0080] In one embodiment, the training process of the defect recognition model includes:

[0081] S201, collecting images and performing preprocessing, including: collecting image data of fire masks with various defects and image data of fire masks without defects, and marking the collected images.

[0082] A high-resolution industrial camera or image acquisition device is used to collect image data of the fire mask, and the collected images include defective fire masks and non-defective fire masks.

[0083] Defects of defective fire masks include: cracks, scratches, dents, particles, bubbles, peeling, color difference, filter element installation defects, etc. Scratches, bubbles, peeling, and color difference are all mask coating defects. Filter element installation defects include: filter elements are not installed properly.

[0084] The captured images are annotated to indicate the location, type and size of the defects.

[0085] After annotation, the image can be cropped, scaled, and normalized to ensure that the image data meets the input requirements of the neural network.

[0086] S202, forward propagation and loss calculation, including: inputting the preprocessed image data into the neural network model, obtaining the prediction result of the model through layer-by-layer calculation, and using the loss function to calculate the difference between the model prediction result and the true label.

[0087] Neural network models can be selected from AlexNet, VGG, ResNet and other models.

[0088] S203, back propagation and parameter update, including: using a back propagation algorithm to update weights and biases in the neural network according to the gradient of the loss function.

[0089] S204, repeat the process of forward propagation, loss calculation, back propagation and parameter update until the performance of the model on the validation set reaches stability.

[0090] Implementation example of intelligent monitoring system for fire mask production process:

[0091] The present invention also provides an intelligent monitoring system for the production process of fire masks. Figure 3 As shown, the fire mask production process intelligent monitoring system includes a processor and a memory, and the memory stores computer program instructions. When the computer program instructions are executed by the processor, the fire mask production process intelligent monitoring method in the fire mask production process intelligent monitoring method embodiment is implemented.

[0092] The intelligent monitoring system for the fire mask production process also includes other components familiar to technical personnel in this field, such as a communication bus and a communication interface. Their settings and functions are known in the art and will not be described in detail here.

[0093] In the present invention, the aforementioned memory may be any tangible medium containing or storing a program that can be used by or in combination with an instruction execution system, apparatus or device. For example, a computer-readable storage medium may be any appropriate magnetic storage medium or magneto-optical storage medium, such as a resistive random access memory RRAM (Resistive Random Access Memory), a dynamic random access memory DRAM (Dynamic Random Access Memory), a static random access memory SRAM (Static Random-Access Memory), an enhanced dynamic random access memory EDRAM (Enhanced Dynamic Random Access Memory), a high-bandwidth memory HBM (High-Bandwidth Memory), a hybrid memory cube HMC (Hybrid Memory Cube), etc., or any other medium that can be used to store the required information and can be accessed by an application, a module, or both. Any such computer storage medium may be part of a device or accessible or connectable to a device. Any application or module described in the present invention may be implemented using computer-readable / executable instructions that may be stored or otherwise maintained by such a computer-readable medium.

[0094] In the description of this specification, "plurality" or "several" means at least two, such as two, three or more, etc., unless otherwise clearly and specifically defined.

[0095] Although this specification has shown and described a number of embodiments of the present invention, it will be apparent to those skilled in the art that such embodiments are provided by way of example only. Those skilled in the art will conceive of many modifications, changes and alternatives without departing from the ideas and spirit of the present invention. It should be understood that in the practice of the present invention, various alternatives to the embodiments of the present invention described herein may be employed.

Claims

1. An intelligent monitoring method for the production process of fire masks, characterized in that: The following steps are involved: Collect the image of the fire mask and grayscale it to obtain a grayscale image of the fire mask; Establishing a window of a preset size with each pixel of the grayscale image as the center; Calculate the noise level of the central pixel of each window. The calculation expression is: ; In the formula, Indicates the number of non-center pixels in the window. Indicates the grayscale deviation of the central pixel in the i-th window. It indicates the grayscale deviation degree of the central pixel in the i-th window from the adjacent pixels in its gradient direction. Indicates the grayscale deviation of the jth non-central pixel in the i-th window. Indicates the grayscale deviation degree of the jth non-central pixel point in the i-th window from the adjacent pixel points in the gradient direction; the grayscale deviation degree is used to characterize the grayscale difference between a certain pixel point and other pixel points in the window; Indicates the noise performance of the central pixel of the i-th window; The bilateral filtering algorithm is used to filter each central pixel point, including: correcting the comprehensive weight corresponding to the central pixel point according to the noise performance degree to obtain a comprehensive weight correction value, wherein the comprehensive weight correction value is negatively correlated with the noise performance degree; the calculation expression of the comprehensive weight correction value is: ； In the formula, represents the comprehensive weight correction value, represents the horizontal coordinate of the center pixel, j represents the vertical coordinate of the center pixel, Indicates the horizontal coordinate of any pixel point in the corresponding window. Indicates the ordinate of any pixel point in the corresponding window. represents the natural exponential function, Represents the gray value of the center pixel, The horizontal axis is , the vertical axis is The gray value of the pixel, The coordinates are ( k , l ) is the normalized noise level of the pixel point, represents the spatial weight parameter, represents the intensity weight parameter; The filtered fire mask grayscale image is input into the preset defect recognition model.

2. The intelligent monitoring method for the production process of fire masks according to claim 1, characterized in that: For the central pixel in a window, the calculation expression of its grayscale deviation is: ； In the formula, It indicates the grayscale difference consistency of the jth non-central pixel in the i-th window before the central pixel is removed. It indicates the grayscale difference consistency of the jth non-central pixel in the i-th window after the central pixel is removed. It represents the change in the grayscale difference consistency of the jth non-central pixel after the central pixel in the window is removed. represents the variance function, represents the variance of the change in the grayscale difference consistency, j=1...n-1, and the grayscale difference consistency is used to characterize the consistency between the grayscale difference of the pixel and the remaining pixels in the window, It represents the grayscale deviation of the central pixel in the i-th window, and n represents the number of pixels in the window.

3. The intelligent monitoring method for the production process of fire masks according to claim 1, characterized in that: For the central pixel in the i-th window, the calculation expression of its grayscale deviation is: ； In the formula, Indicates the chaos degree of the pixels in the i-th window, It indicates the grayscale difference consistency of the jth non-central pixel in the i-th window before the central pixel is removed. It indicates the grayscale difference consistency of the jth non-central pixel in the i-th window after the central pixel is removed. It represents the change in the grayscale difference consistency of the jth non-central pixel after the central pixel in the window is removed. represents the variance function, The variance of the change in the grayscale difference consistency is represented, j=1...n-1, and the chaos degree is used to characterize the grayscale value difference between different pixels in the window. It represents the grayscale deviation of the central pixel in the i-th window, and n represents the number of pixels in the window.

4. The intelligent monitoring method for the production process of fire masks according to claim 3 is characterized in that: For a certain pixel point in a certain window, the calculation method of its grayscale difference consistency includes: respectively calculating the difference between the grayscale value of the pixel point and the grayscale values ​​of each of the other pixel points in the window, and then calculating the variance of the difference, and taking the inverse of the variance of the difference as the grayscale difference consistency of the pixel point.

5. The intelligent monitoring method for the production process of fire masks according to claim 3 is characterized in that: The calculation expression of the chaos degree of pixels in a certain window is: ； In the formula, Indicates the chaos degree of the pixels in the i-th window, Indicates the number of pixels in the window. Represents the gray value of the central pixel in the i-th window, Represents the gray value of the j-th pixel in the i-th window.

6. The intelligent monitoring method for the production process of fire masks according to claim 3, characterized in that: The calculation expression of the chaos degree of pixels in a certain window is: ； In the formula, Represents the variance of the grayscale value of the pixel in the i-th window, Indicates the chaos degree of the pixels in the i-th window, Indicates the number of pixels in the window. Represents the gray value of the central pixel in the i-th window, Represents the gray value of the j-th pixel in the i-th window.

7. The intelligent monitoring method for the production process of fire masks according to any one of claims 1 to 6, characterized in that: The size of the window is 5 5.

8. An intelligent monitoring system for the production process of fire masks, characterized in that: It includes a processor and a memory, the memory stores computer program instructions, and when the computer program instructions are executed by the processor, the intelligent monitoring method for the fire mask production process described in any one of claims 1 to 7 is implemented.

Images