An intelligent detection system and method for wall panel defects based on image recognition

Abstract

The present invention discloses an intelligent detection system and method for wall panel defects based on image recognition, which relates to the technical field of image processing, and includes an image acquisition module, an image detection module, and a result fusion and output module; the present invention ensures data quality through the image acquisition and preprocessing module, and uses two convolutional neural network models to quickly and accurately detect damage and color difference defects respectively, and then through the result fusion and output module, further comprehensively analyze through intelligent algorithms to eliminate false detections and ensure the comprehensiveness and accuracy of the detection results. This method not only improves the detection efficiency and accuracy, but also reduces the labor cost, is applicable to various types of defect detections, and significantly improves the quality control standard of wall panel products.

Description

Technical Field

[0001] The present invention relates to the technical field of image processing, and specifically to an intelligent detection system and method for wall panel defects based on image recognition. Background Art

[0002] With the improvement of people's living standards, more and more people begin to pursue a high-quality life and have higher and higher requirements for the home environment. When decorating a house, some high-end decoration materials will be purchased to enhance the overall texture of the room. For example, wall panels have good sound insulation effects, can reduce noise pollution, and have good heat insulation performance, which can reduce heat loss. In addition, wall panels also have good fire prevention and flame retardant functions, with a higher safety factor. However, in the actual production process, due to process problems or human factors, various defects often occur, affecting the aesthetics of the product and reducing the product quality. Therefore, quality control is required.

[0003] Currently, the traditional detection of wall panel defects mainly relies on manual visual inspection. However, this inspection method has low efficiency, low accuracy, and is prone to problems such as missed detection. To improve this situation, in recent years, some researchers have tried to use computer vision technology for wall panel defect detection. The advantage of this method is high speed and high precision, but it can only detect specific types of products and cannot detect multiple types of defects simultaneously. Summary of the Invention

[0004] The present invention provides an intelligent detection system and method for wall panel defects based on image recognition to solve the above problems existing in the prior art.

[0005] According to one aspect of the present invention, an intelligent detection system for wall panel defects based on image recognition is provided, including an image acquisition module, an image detection module, and a result fusion and output module.

[0006] The image acquisition module is used to obtain the wall panel image to be detected, preprocess the wall panel image data to be detected, and obtain the preprocessed image data. The image preprocessing operations include image grayscale conversion, filtering, and size normalization, etc.

[0007] The image detection module: detects surface breakage and color difference defects of the wall panel through a pre-trained first convolutional neural network model and a second convolutional neural network model; the image detection module includes a breakage detection unit and a color difference defect detection unit.

[0008] The breakage detection unit: performs wall panel breakage detection on the preprocessed image data through a pre-trained first convolutional neural network model to obtain a first detection result.

[0009] The color difference defect detection unit: performs wall panel color difference defect detection on the preprocessed image data through a pre-trained second convolutional neural network model to obtain a second detection result;

[0010] The result fusion and output module is used to fuse the first detection result and the second detection result and output the final detection result.

[0011] Optionally, the image acquisition module is further used to set the position and angle of the camera, place the camera at a position at a predetermined height from the wall panel to be measured; adjust the angle of the camera so that the center of the camera lens forms a predetermined included angle with the horizontal direction of the wall panel to be measured.

[0012] Optionally, the damage detection unit includes: performing forward propagation on the preprocessed image data using the first convolutional neural network model to obtain a first prediction box and a first confidence score; if the first confidence score is greater than or equal to a predetermined threshold, then determining the first prediction box as the first detection result;

[0013] The color difference defect detection unit includes: performing forward propagation on the preprocessed image data using the second convolutional neural network model to obtain a second prediction box and a second confidence score; if the second confidence score is greater than or equal to a predetermined threshold, then determining the second prediction box as the second detection result.

[0014] Optionally, performing forward propagation on the preprocessed image data using the first convolutional neural network model further includes a feature extraction network, a Region Proposal Network (RPN), and a ROI pooling layer.

[0015] Among them, the feature extraction network is used to extract a set of feature maps from the preprocessed image data;

[0016] The Region Proposal Network (RPN) is used to predict multiple region proposal boxes according to the feature maps;

[0017] The ROI pooling layer is used to perform convolutional operations on each region proposal box respectively, and then compress the information within the region proposal box into a feature map of a fixed size.

[0018] Optionally, performing forward propagation on the preprocessed image data using the second convolutional neural network model further includes sub-pixel gray matrix construction, determination of subordinate main pixel points, gray weight calculation, determination of sub-pixel point gray values, and sub-pixel image construction.

[0019] Among them, the sub-pixel gray matrix construction is used to convert the gray image of the preprocessed image data into a sub-pixel gray matrix, and the sub-pixel gray matrix includes main pixel points and sub-pixel points;

[0020] The affiliated main pixel point determination is used to calculate the gray - level difference and distance difference between the sub - pixel point and the candidate main pixel point, and determine the candidate main pixel point with the highest membership degree corresponding to the sub - pixel point according to a preset threshold, which is the main pixel point to which the sub - pixel point belongs;

[0021] The gray - level weight calculation is used to calculate the gray - level weight of the sub - pixel point according to the gray - level value and position of the main pixel point to which the sub - pixel point belongs;

[0022] The sub - pixel point gray - level value determination is used to determine the gray - level value of the sub - pixel point according to the gray - level value of the main pixel point to which the sub - pixel point belongs, the gray - level weight of the sub - pixel point, and the gray - level value of the main pixel point;

[0023] The sub - pixel image construction is used to construct a sub - pixel image according to the gray - level value of the sub - pixel point.

[0024] Optionally, the affiliated main pixel point determination is specifically used to calculate the gray - level difference and distance difference between the sub - pixel point and the candidate main pixel point, and determine the main pixel point to which the sub - pixel point belongs according to a preset threshold, including:

[0025] If |G - G″| ≤ ε1 and d ≤ ε2, then the sub - pixel point belongs to the candidate main pixel point, where G represents the gray - level value of the sub - pixel point, G″ represents the gray - level value of the candidate main pixel point, d represents the distance between the sub - pixel point and the candidate main pixel point, ε1 represents the gray - level difference threshold, and ε2 represents the distance difference threshold.

[0026] Optionally, the gray - level weight calculation is specifically used to calculate the gray - level weight of the sub - pixel point according to the gray - level value and position of the main pixel point to which the sub - pixel point belongs, including:

[0027] If the sub - pixel point belongs to the main pixel point in the i - th row and j - th column, then the gray - level weight of the sub - pixel point is w=(i - α)·(β - j), where α represents the proportion of the row number of the main pixel point to which the sub - pixel point belongs in the total number of rows, and β represents the proportion of the column number of the main pixel point to which the sub - pixel point belongs in the total number of columns.

[0028] Optionally, the sub - pixel point gray - level value determination is specifically used to determine the gray - level value of the sub - pixel point according to the gray - level value of the main pixel point to which the sub - pixel point belongs, the gray - level weight of the sub - pixel point, and the gray - level value of the main pixel point, including:

[0029] If the sub-pixel point belongs to the main pixel point in the i-th row and j-th column, the gray value of the sub-pixel point is g = w * gi + (1 - w) * gj, where g represents the gray value of the sub-pixel point, w represents the gray weight of the sub-pixel point, gi represents the gray value of the main pixel point in the i-th row, and gj represents the gray value of the main pixel point in the j-th column.

[0030] Optionally, the result fusion and output module is used to fuse the first detection result and the second detection result and output the final detection result, including:

[0031] Obtaining the position information of the first type of defect in the wall panel image according to the first detection result; obtaining the position information of the second type of defect according to the second detection result; the first type of defect is a hole, crack or missing block, and the second type of defect is a color difference defect;

[0032] Wherein, when the position areas of the first detection result and the second detection result coincide, the coincident area is further analyzed. When the coincidence result is a misdetection caused by color difference, the marked area position is marked as damaged; when there is no coincidence in the position areas of the first detection result and the second detection result, it is determined that both results are valid defect detection results, and the results are respectively recorded, that is, the final detection result is obtained.

[0033] According to another aspect of the present invention, there is provided an intelligent detection method for wall panel defects based on image recognition, including the following steps:

[0034] Obtaining a wall panel image to be detected through a camera;

[0035] Preprocessing the wall panel image data to be detected to obtain preprocessed image data;

[0036] Performing wall panel damage detection on the preprocessed image data through a pre-trained first convolutional neural network model to obtain a first detection result;

[0037] Performing wall panel color difference defect detection on the preprocessed image data through a pre-trained second convolutional neural network model to obtain a second detection result;

[0038] Fusing the first detection result and the second detection result and outputting the final detection result;

[0039] Wherein, performing forward propagation on the preprocessed image data by using the first convolutional neural network model to obtain a first prediction box and a first confidence score; if the first confidence score is greater than or equal to a predetermined threshold, determining the first prediction box as the first detection result;

[0040] Perform forward propagation on the preprocessed image data using the second convolutional neural network model to obtain a second prediction box and a second confidence score; if the second confidence score is greater than or equal to a predetermined threshold, determine the second prediction box as the second detection result;

[0041] Fuse the first detection result and the second detection result and output the final detection result, including:

[0042] Obtain the position information of the first type of defect in the wall panel image according to the first detection result; obtain the position information of the second type of defect according to the second detection result; the first type of defect is a hole, crack or missing block, and the second type of defect is a color difference defect;

[0043] Wherein, when the position areas of the first detection result and the second detection result overlap, analyze the overlapping area again. When the overlapping result is a misdetection caused by color difference, mark the area position as damaged; when there is no overlap in the position areas of the first detection result and the second detection result, it is determined that both results are valid defect detection results, record the results respectively, and thus obtain the final detection result.

[0044] The beneficial effects of the present invention are:

[0045] Efficiently obtain wall panel images through the automated image acquisition module and perform necessary preprocessing on them, including steps such as grayscale conversion, filtering, and size standardization. The preprocessing operation not only ensures the quality of the image data but also lays a solid foundation for subsequent defect detection, reduces manual operations, and significantly improves the efficiency of the detection process. At the same time, using a pre-trained convolutional neural network model, rapid detection of damage and color difference defects in wall panel images is achieved;

[0046] The present invention uses two dedicated convolutional neural network models to accurately identify different types of defects. The damage detection model adopts a feature extraction network, a region proposal network RPN, and ROI pooling layer technology to optimize the positioning process of the damaged area. The color difference defect model realizes sub-pixel level detection of color difference through the construction of a sub-pixel gray matrix and precise calculation of gray weights. The combined use of these two models not only improves the accuracy and reliability of detection but also reduces the possibility of missed detection and misdetection;

[0047] Through the result fusion and output module, intelligent algorithms are used to comprehensively analyze the detection results of damage and color difference defects, re-analyze the possibly overlapping defect areas, eliminate false detections, ensuring the comprehensiveness and accuracy of the detection results. It can not only detect single-type defects, but also simultaneously identify multiple defects including holes, cracks, missing blocks, and color differences, greatly expanding the application scope of detection, ensuring high standards for the quality of wall panel products. Through this comprehensive method, the automation, precision, and intelligence of wall panel defect detection are achieved, effectively improving the detection efficiency and accuracy while reducing labor costs. Description of the Drawings

[0048] For better understanding and implementation, the technical solutions of this application will be described in detail below with reference to the drawings.

[0049] Figure 1 Schematic diagram of the structure of an intelligent wall panel defect detection system based on image recognition provided in the first embodiment of this application;

[0050] Figure 2 Schematic diagram of the process of forward propagation of the preprocessed image data by using the first convolutional neural network model in an intelligent wall panel defect detection system based on image recognition provided in the first embodiment of this application;

[0051] Figure 3 Schematic diagram of the process of forward propagation of the preprocessed image data by using the second convolutional neural network model in an intelligent wall panel defect detection system based on image recognition provided in the first embodiment of this application;

[0052] Figure 4 Schematic diagram of the process of an intelligent wall panel defect detection method based on image recognition provided in the second embodiment of this application. Detailed Embodiments

[0053] To further elaborate on the technical means and effects adopted by the present invention to achieve the predetermined invention purpose, exemplary embodiments will be described in detail here, and their examples are shown in the drawings. When the following description refers to the drawings, unless otherwise indicated, the same numbers in different drawings represent the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with this application. On the contrary, they are merely examples of methods and systems consistent with some aspects of this application as detailed in the appended claims.

[0054] The terms used in this application are for the purpose of describing specific embodiments only and are not intended to limit this application. The singular forms "a", "the", and "said" used in this application and the appended claims are also intended to include the plural forms unless the context clearly dictates otherwise. It should also be understood that the term "and / or" as used herein refers to and encompasses any and all possible combinations of one or more of the associated listed items.

[0055] The following, in conjunction with the accompanying drawings and preferred embodiments, details the specific implementation manners, features, and effects of the present invention.

[0056] Embodiment 1

[0057] Referring to Figure 1 , an intelligent detection system for wall panel defects based on image recognition provided by an embodiment of the present invention includes an image acquisition module, an image detection module, and a result fusion and output module, wherein:

[0058] Image acquisition module: used to obtain the wall panel image to be detected, preprocess the wall panel image data to be detected, and obtain the preprocessed image data. Image preprocessing operations include image grayscale conversion, filtering, and size normalization, etc.;

[0059] Image detection module: detects surface breakage and color difference defects of the wall panel through a pre-trained first convolutional neural network model and a second convolutional neural network model; the image detection module includes a breakage detection unit and a color difference defect detection unit,

[0060] The breakage detection unit: performs wall panel breakage detection on the preprocessed image data through a pre-trained first convolutional neural network model to obtain a first detection result;

[0061] The color difference defect detection unit: performs wall panel color difference defect detection on the preprocessed image data through a pre-trained second convolutional neural network model to obtain a second detection result;

[0062] Result fusion and output module: used to fuse the first detection result and the second detection result and output the final detection result.

[0063] The intelligent detection system for wall panel defects based on image recognition provided by this embodiment analyzes the image of the wall panel, automatically identifies the existing defects therein, without manual participation, greatly improves the detection efficiency, reduces the labor cost at the same time, and can also be applied to the defect detection of other types of boards, with a very wide range of applications.

[0064] Of course, as a preferred embodiment, the image acquisition module is further configured to set the position and angle of the camera, place the camera at a position at a predetermined height from the wall panel to be measured; adjust the angle of the camera so that the center of the camera lens forms a predetermined angle with the horizontal direction of the wall panel to be measured. The advantage of doing this is that it can ensure that the captured pictures are clear and complete, which is beneficial for subsequent image processing and analysis.

[0065] Preferably, the damage detection unit includes: performing forward propagation on the preprocessed image data by using the first convolutional neural network model to obtain a first prediction box and a first confidence score; if the first confidence score is greater than or equal to a predetermined threshold, determining the first prediction box as a first detection result.

[0066] The color difference defect detection unit includes: performing forward propagation on the preprocessed image data by using the second convolutional neural network model to obtain a second prediction box and a second confidence score; if the second confidence score is greater than or equal to a predetermined threshold, determining the second prediction box as a second detection result.

[0067] Preferably, as Figure 2 shown, performing forward propagation on the preprocessed image data by using the first convolutional neural network model further includes:

[0068] S11. Feature extraction network: used to extract a set of feature maps from the preprocessed image data.

[0069] S12. Region Proposal Network RPN: used to predict a plurality of region proposal boxes according to the feature maps.

[0070] S13. ROI pooling layer: used to perform convolution operations on each region proposal box respectively, and then compress the information within the region proposal box into a feature map of a fixed size.

[0071] The intelligent wall panel defect detection system based on image recognition provided in this embodiment extracts a set of feature maps from the input image through the feature extraction network, then predicts a plurality of region proposal boxes that may have problems through the Region Proposal Network RPN, and finally compresses the information within these region proposal boxes into a feature map of a fixed size through the ROI pooling layer, so as to realize the rapid positioning and recognition of wall panel defects.

[0072] Preferably, as Figure 3 shown, performing forward propagation on the preprocessed image data by using the second convolutional neural network model further includes:

[0073] S21. Sub-pixel gray matrix construction: used to convert the gray image of the preprocessed image data into a sub-pixel gray matrix, and the sub-pixel gray matrix includes main pixel points and sub-pixel points.

[0074] S22, Determination of the affiliated main pixel point: used to calculate the gray - level difference and distance difference between the sub - pixel point and the candidate main pixel point, and determine the main pixel point to which the sub - pixel point belongs according to a preset threshold;

[0075] S23, Gray - level weight calculation: used to calculate the gray - level weight of the sub - pixel point according to the gray - level value and position of the main pixel point to which the sub - pixel point belongs;

[0076] S24, Determination of the gray - level value of the sub - pixel point: used to determine the gray - level value of the sub - pixel point according to the gray - level value of the main pixel point to which the sub - pixel point belongs, the gray - level weight of the sub - pixel point, and the gray - level value of the main pixel point;

[0077] S25, Sub - pixel image construction: used to construct a sub - pixel image according to the gray - level value of the sub - pixel point.

[0078] The intelligent wall - panel defect detection system based on image recognition provided in this embodiment converts the gray - level image of the pre - processed image data into a sub - pixel gray - level matrix, and then determines the main pixel point to which the sub - pixel point belongs according to the gray - level difference and distance difference between the sub - pixel point and the candidate main pixel point, so as to achieve accurate recognition of wall - panel color - difference defects.

[0079] Further, the determination of the affiliated main pixel point is specifically used to calculate the gray - level difference and distance difference between the sub - pixel point and the candidate main pixel point, and determine the main pixel point to which the sub - pixel point belongs according to a preset threshold, including:

[0080] If |G - G″| ≤ ε1 and d ≤ ε2, then the sub - pixel point belongs to the candidate main pixel point, where G represents the gray - level value of the sub - pixel point, G″ represents the gray - level value of the candidate main pixel point, d represents the distance between the sub - pixel point and the candidate main pixel point, ε1 represents the gray - level difference threshold, and ε2 represents the distance - difference threshold.

[0081] Further, the gray - level weight calculation is specifically used to calculate the gray - level weight of the sub - pixel point according to the gray - level value and position of the main pixel point to which the sub - pixel point belongs, including:

[0082] If the sub - pixel point belongs to the main pixel point in the i - th row and j - th column, then the gray - level weight of the sub - pixel point is w=(i - α)·(β - j), where α represents the proportion of the row number of the main pixel point to which the sub - pixel point belongs in the total number of rows, and β represents the proportion of the column number of the main pixel point to which the sub - pixel point belongs in the total number of columns.

[0083] Further, the determination of the sub-pixel point gray value is specifically used to determine the gray value of the sub-pixel point according to the gray value of the main pixel point to which the sub-pixel point belongs, the gray weight of the sub-pixel point, and the gray value of the main pixel point, including:

[0084] If the sub-pixel point belongs to the main pixel point in the i-th row and j-th column, the gray value of the sub-pixel point is g = w * gi + (1 - w) * gj, where g represents the gray value of the sub-pixel point, w represents the gray weight of the sub-pixel point, gi represents the gray value of the main pixel point in the i-th row, and gj represents the gray value of the main pixel point in the j-th column.

[0085] It should be noted that through the advanced damage detection unit and color difference defect detection unit, the rapid and accurate identification of wall panel defects is realized. The damage detection unit uses the feature extraction network, region proposal network RPN, and ROI pooling layer to compress the information within these boxes into a feature map of a fixed size, so as to efficiently locate and identify potential damaged areas from the preprocessed image. The color difference defect detection unit realizes the accurate identification of the color difference defects of the wall panel by constructing a sub-pixel gray matrix, accurately calculating the gray weight and sub-pixel point gray value, ensuring the accurate measurement and analysis of the color difference at the sub-pixel level. The system further adopts the determination of the subordinate main pixel points and the calculation of the gray weight, enhancing the recognition ability of sub-pixel level defects. Finally, the result fusion and output module synthesizes the results of the two detection units, intelligently eliminates false detections and accurately fuses the effective defect information, and outputs an accurate and reliable final detection report, providing strong technical support for the quality control of wall panels. An intelligent wall panel defect detection system based on image recognition significantly improves the automation level and accuracy of defect detection, ensuring the high-standard quality of wall panel products.

[0086] Preferably, the result fusion and output module is used to fuse the first detection result and the second detection result and output the final detection result, including:

[0087] Obtain the position information of the first type of defect in the wall panel image according to the first detection result; obtain the position information of the second type of defect according to the second detection result; the first type of defect is a hole, crack or missing block, and the second type of defect is a color difference defect;

[0088] Among them, when the position areas of the first detection result and the second detection result coincide, analyze the overlapping area again. When the overlapping result is a false detection caused by color difference, mark the area position as damaged; when there is no overlapping position area between the first detection result and the second detection result, it is determined that both results are valid defect detection results, record the results respectively, and the final detection result is obtained.

[0089] In summary, the method of the present invention realizes the automatic detection of wall panels, improves the detection efficiency, and reduces the labor cost; the method of the present invention adopts a deep learning algorithm to improve the detection accuracy; the method of the present invention can detect various types of defects simultaneously, with a wide range of applications; the method of the present invention takes into account the correlation between different defects and avoids misjudgment.

[0090] Embodiment 2

[0091] Refer to Figure 4 , a method for intelligent detection of wall panel defects based on image recognition according to an embodiment of the present invention, the method comprising the following steps:

[0092] S100: Obtain a wall panel image to be detected through a camera;

[0093] S200: Preprocess the wall panel image data to be detected to obtain preprocessed image data, and the image preprocessing operations include image grayscaling, filtering, and size normalization, etc.;

[0094] S300: Perform wall panel breakage detection on the preprocessed image data through a pre-trained first convolutional neural network model to obtain a first detection result;

[0095] S400: Perform wall panel color difference defect detection on the preprocessed image data through a pre-trained second convolutional neural network model to obtain a second detection result;

[0096] S500: Fuse the first detection result and the second detection result and output a final detection result.

[0097] Specifically, for the method for intelligent detection of wall panel defects based on image recognition provided in this embodiment, a wall panel image to be detected is obtained through a camera, the wall panel image data to be detected is preprocessed to obtain preprocessed image data, wall panel breakage detection is performed on the preprocessed image data through a pre-trained first convolutional neural network model to obtain a first detection result, wall panel color difference defect detection is performed on the preprocessed image data through a pre-trained second convolutional neural network model to obtain a second detection result, the first detection result and the second detection result are fused, and a final detection result is output. The method improves the detection efficiency and reduces the labor cost; adopts a deep learning algorithm to improve the detection accuracy; can detect various types of defects simultaneously, with a wide range of applications; takes into account the correlation between different defects and avoids misjudgment.

[0098] Preferably, the method further includes:

[0099] Set the position and angle of the camera, and place the camera at a position at a predetermined height from the wall panel to be measured; adjust the angle of the camera so that the center of the camera lens forms a predetermined angle with the horizontal direction of the wall panel to be measured.

[0100] Preferably, according to step S300, it further includes: performing forward propagation on the preprocessed image data by using the first convolutional neural network model to obtain a first prediction box and a first confidence score; if the first confidence score is greater than or equal to a predetermined threshold, determining the first prediction box as the first detection result.

[0101] According to step S400, it further includes: performing forward propagation on the preprocessed image data by using the second convolutional neural network model to obtain a second prediction box and a second confidence score; if the second confidence score is greater than or equal to a predetermined threshold, determining the second prediction box as the second detection result.

[0102] Preferably, performing forward propagation on the preprocessed image data by using the first convolutional neural network model includes:

[0103] Extracting a set of feature maps from the preprocessed image data;

[0104] Predicting multiple region proposal boxes based on the feature maps;

[0105] After performing convolution operations on each region proposal box respectively, compressing the information within the region proposal box into a feature map of a fixed size.

[0106] Preferably, performing forward propagation on the preprocessed image data by using the second convolutional neural network model includes:

[0107] Converting the grayscale image of the preprocessed image data into a sub-pixel grayscale matrix, where the sub-pixel grayscale matrix includes main pixel points and sub-pixel points;

[0108] Calculating the grayscale difference and distance difference between the sub-pixel points and candidate main pixel points, and determining the main pixel point to which the sub-pixel points belong according to a preset threshold;

[0109] Calculating the grayscale weight of the sub-pixel points according to the grayscale value and position of the main pixel point to which the sub-pixel points belong;

[0110] Determining the grayscale value of the sub-pixel points according to the grayscale value of the main pixel point to which the sub-pixel points belong, the grayscale weight of the sub-pixel points, and the grayscale value of the main pixel point;

[0111] Constructing a sub-pixel image according to the grayscale value of the sub-pixel points.

[0112] Preferably, calculating the gray - level difference and distance difference between the sub - pixel point and the candidate main pixel point, and determining the main pixel point to which the sub - pixel point belongs according to a preset threshold value, includes:

[0113] If |G - G″| ≤ ε1 and d ≤ ε2, then the sub - pixel point belongs to the candidate main pixel point, where G represents the gray - level value of the sub - pixel point, G″ represents the gray - level value of the candidate main pixel point, d represents the distance between the sub - pixel point and the candidate main pixel point, ε1 represents the gray - level difference threshold value, and ε2 represents the distance difference threshold value.

[0114] Preferably, calculating the gray - level weight of the sub - pixel point according to the gray - level value and position of the main pixel point to which the sub - pixel point belongs, includes:

[0115] If the sub - pixel point belongs to the main pixel point in the i - th row and j - th column, then the gray - level weight of the sub - pixel point is w=(i - α)·(β - j), where α represents the proportion of the row number of the main pixel point to which the sub - pixel point belongs in the total number of rows, and β represents the proportion of the column number of the main pixel point to which the sub - pixel point belongs in the total number of columns.

[0116] Preferably, determining the gray - level value of the sub - pixel point according to the gray - level value of the main pixel point to which the sub - pixel point belongs, the gray - level weight of the sub - pixel point, and the gray - level value of the main pixel point, includes:

[0117] If the sub - pixel point belongs to the main pixel point in the i - th row and j - th column, then the gray - level value of the sub - pixel point is g = w*gi+(1 - w)*gj, where g represents the gray - level value of the sub - pixel point, w represents the gray - level weight of the sub - pixel point, gi represents the gray - level value of the main pixel point in the i - th row, and gj represents the gray - level value of the main pixel point in the j - th column.

[0118] Preferably, fusing the first detection result and the second detection result and outputting the final detection result, includes:

[0119] Obtaining the position information of the first - type defects in the wall panel image according to the first detection result; obtaining the position information of the second - type defects according to the second detection result; the first - type defects are holes, cracks or missing blocks, and the second - type defects are color - difference defects;

[0120] Among them, when the position areas of the first detection result and the second detection result overlap, then analyze the overlapping area. When the overlapping result is a mis - detection caused by color - difference, then mark the area position as damaged; when the position areas of the first detection result and the second detection result do not overlap, then consider both results as valid defect detection results, record the results respectively, that is, obtain the final detection result.

[0121] Embodiment III

[0122] After determining the gray value of the sub-pixel point, this embodiment further includes: removing outliers; performing interpolation operations on the remaining sub-pixel points to obtain the final gray value of the sub-pixel point, specifically including:

[0123] Determine the gray value and position: Determine that the gray values of pixel points A and B are g A and g B , and their positions in the image;

[0124] Calculate the distance: Calculate the distance d AB between pixel points A and B. If A and B are adjacent pixels, this distance is usually 1;

[0125] Determine the specific position of point P relative to A, and calculate the distance d P from point P to A. If P is located exactly in the middle of A and B, then d P will be d AB / 2;

[0126] Apply the linear interpolation formula: Use the linear interpolation formula to calculate the gray value g P of point P,

[0127]

[0128] If point P has a gray weight w, which is calculated based on the gray value and position of the main pixel point to which point P belongs, then the final gray value g′ P of point P is expressed as:

[0129] g′ P = w × g P +(1 - w) × (g A + g B ) / 2

[0130] During the calculation process, if point P is an outlier, remove or correct its influence according to the method of determining the gray value and position;

[0131] Repeat the calculation: For all sub-pixel points that need to perform interpolation operations, repeat the above steps;

[0132] Construct the final sub-pixel image: Place the gray values g′ P of all calculated sub-pixel points back to their correct positions in the image to construct the final sub-pixel image.

[0133] Through the above steps, the interpolation operation of sub-pixel points in the wall panel image can be effectively performed, thereby improving the accuracy and resolution of color difference detection. In the detection of wall panel defects, very small defects can be identified.

[0134] The above are only the preferred embodiments of the present invention and do not impose any form of limitation on the present invention. Although the present invention has been disclosed above in the preferred embodiments, it is not intended to limit the present invention. Any person skilled in the art can make some changes or modifications to equivalent embodiments by using the above-disclosed technical content within the scope of the technical solution of the present invention. However, as long as it does not depart from the content of the technical solution of the present invention, any brief modifications, equivalent changes, and modifications made to the above embodiments based on the technical essence of the present invention still fall within the scope of the technical solution of the present invention.

Claims

1. An intelligent detection system for wall panel defects based on image recognition, characterized in that: It includes an image acquisition module, an image detection module, and a result fusion and output module. The image acquisition module is used to obtain the wall panel image to be detected, preprocess the wall panel image data to be detected, and obtain the preprocessed image data. The image detection module: detects the surface breakage and color difference defects of the wall panel through a pre-trained first convolutional neural network model and a second convolutional neural network model; the image detection module includes a breakage detection unit and a color difference defect detection unit. The breakage detection unit: performs wall panel breakage detection on the preprocessed image data through a pre-trained first convolutional neural network model to obtain a first detection result. The color difference defect detection unit: performs wall panel color difference defect detection on the preprocessed image data through a pre-trained second convolutional neural network model to obtain a second detection result. The result fusion and output module is used to fuse the first detection result and the second detection result and output the final detection result, specifically including: Obtaining the position information of the first type of defect in the wall panel image according to the first detection result; obtaining the position information of the second type of defect according to the second detection result; the first type of defect is a hole, crack or missing block, and the second type of defect is a color difference defect. Among them, when the position areas of the first detection result and the second detection result overlap, analyze the overlapping area again. When the overlapping result is a misdetection caused by color difference, mark the area position as breakage; when there is no overlap in the position areas of the first detection result and the second detection result, it is determined that both results are valid defect detection results, record the results respectively, and obtain the final detection result. The image acquisition module is also used to set the position and angle of the camera, place the camera at a position with a predetermined height from the wall panel to be measured; adjust the angle of the camera so that the center of the camera lens forms a predetermined angle with the horizontal direction of the wall panel to be measured.

2. The intelligent detection system for wall panel defects based on image recognition according to claim 1, wherein: The breakage detection unit includes: performing forward propagation on the preprocessed image data using the first convolutional neural network model to obtain a first prediction box and a first confidence score; if the first confidence score is greater than or equal to a predetermined threshold, determine the first prediction box as the first detection result. The color difference defect detection unit includes: performing forward propagation on the preprocessed image data using the second convolutional neural network model to obtain a second prediction box and a second confidence score; if the second confidence score is greater than or equal to a predetermined threshold, determine the second prediction box as the second detection result.

3. An intelligent detection system for wall panel defects based on image recognition according to claim 2, characterized in that: Performing forward propagation on the preprocessed image data using the first convolutional neural network model further includes a feature extraction network, a Region Proposal Network (RPN), and an ROI pooling layer. Among them, the feature extraction network is used to extract a set of feature maps from the preprocessed image data. The Region Proposal Network (RPN) is used to predict multiple region proposal boxes according to the feature maps. The ROI pooling layer is used to perform convolution operations on each region proposal box respectively, and then compress the information within the region proposal box into a feature map with a fixed size.

4. An intelligent detection system for wall panel defects based on image recognition according to claim 2, characterized in that: Performing forward propagation on the preprocessed image data using the second convolutional neural network model further includes sub-pixel gray matrix construction, determination of the belonging main pixel point, gray weight calculation, determination of the gray value of the sub-pixel point, and sub-pixel image construction. Among them, the sub-pixel gray matrix construction is used to convert the gray image of the preprocessed image data into a sub-pixel gray matrix, and the sub-pixel gray matrix includes main pixel points and sub-pixel points. The determination of the belonging main pixel point is used to calculate the gray difference and distance difference between the sub-pixel point and the candidate main pixel point, and according to a preset threshold, determine the candidate main pixel point with the highest membership degree corresponding to the sub-pixel point, that is, the main pixel point to which the sub-pixel point belongs. The gray weight calculation is used to calculate the gray weight of the sub-pixel point according to the gray value and position of the main pixel point to which the sub-pixel point belongs. The determination of the gray value of the sub-pixel point is used to determine the gray value of the sub-pixel point according to the gray value of the main pixel point to which the sub-pixel point belongs, the gray weight of the sub-pixel point, and the gray value of the main pixel point. The sub-pixel image construction is used to construct a sub-pixel image according to the gray value of the sub-pixel point.

5. The intelligent detection system for wall panel defects based on image recognition according to claim 4, characterized in that: The determination of the belonging main pixel point is specifically used to calculate the gray difference and distance difference between the sub-pixel point and the candidate main pixel point, and according to a preset threshold, determine the main pixel point to which the sub-pixel point belongs, including: If |G - G″| ≤ ε1 and d ≤ ε2, then the sub-pixel point belongs to the candidate main pixel point, where G represents the gray value of the sub-pixel point, G″ represents the gray value of the candidate main pixel point, d represents the distance between the sub-pixel point and the candidate main pixel point, ε1 represents the gray difference threshold, and ε2 represents the distance difference threshold.

6. The intelligent detection system for wall panel defects based on image recognition according to claim 4, characterized in that: The gray weight calculation is specifically used to calculate the gray weight of the sub-pixel point according to the gray value and position of the main pixel point to which the sub-pixel point belongs, including: If the sub-pixel point belongs to the main pixel point in the i-th row and j-th column, then the gray weight of the sub-pixel point is w = (i - α)·(β - j), where α represents the proportion of the row number of the main pixel point to which the sub-pixel point belongs in the total number of rows, and β represents the proportion of the column number of the main pixel point to which the sub-pixel point belongs in the total number of columns.

7. An intelligent detection system for wall panel defects based on image recognition according to claim 4, characterized in that: The determination of the gray value of the sub-pixel point is specifically used to determine the gray value of the sub-pixel point according to the gray value of the main pixel point to which the sub-pixel point belongs, the gray weight of the sub-pixel point, and the gray value of the main pixel point, including: If the sub-pixel point belongs to the main pixel point in the i-th row and j-th column, then the gray value of the sub-pixel point is g = w*gi + (1 - w)*gj, where g represents the gray value of the sub-pixel point, w represents the gray weight of the sub-pixel point, gi represents the gray value of the main pixel point in the i-th row, and gj represents the gray value of the main pixel point in the j-th column.

8. An intelligent detection method for wall panel defects based on image recognition, applied to an intelligent detection system for wall panel defects based on image recognition according to any one of claims 1-7, characterized in that: Including the following steps: Obtaining a wall panel image to be detected through a camera; Preprocessing the wall panel image data to be detected to obtain preprocessed image data. Perform wall panel damage detection on the preprocessed image data through a pre-trained first convolutional neural network model to obtain a first detection result; Perform wall panel color difference defect detection on the preprocessed image data through a pre-trained second convolutional neural network model to obtain a second detection result; Fuse the first detection result and the second detection result and output the final detection result; Among them, use the first convolutional neural network model to perform forward propagation on the preprocessed image data to obtain a first prediction box and a first confidence score; if the first confidence score is greater than or equal to a predetermined threshold, determine the first prediction box as the first detection result; Use the second convolutional neural network model to perform forward propagation on the preprocessed image data to obtain a second prediction box and a second confidence score; if the second confidence score is greater than or equal to a predetermined threshold, determine the second prediction box as the second detection result; The fusing the first detection result and the second detection result and outputting the final detection result includes: Obtain the position information of the first type of defect in the wall panel image according to the first detection result; obtain the position information of the second type of defect according to the second detection result; the first type of defect is a hole, crack or missing block, and the second type of defect is a color difference defect; Among them, when the position areas of the first detection result and the second detection result overlap, analyze the overlapping area again. When the overlapping result is a misdetection caused by color difference, mark the area position as damaged; when there is no overlap in the position areas of the first detection result and the second detection result, it is determined that both results are valid defect detection results, record the results separately, and obtain the final detection result.

Images