An asphalt concrete core wall quality detection method based on multi-modal fusion

Abstract

This invention discloses a method for quality inspection of asphalt concrete core walls based on multimodal fusion, belonging to the field of image processing technology. The method includes the following steps: S1, acquiring sample images of the area to be inspected in the asphalt concrete core wall; S2, determining the local update value of each pixel in the sample image based on its grayscale value and corresponding neighborhood fitness; S3, obtaining the grayscale co-occurrence matrix of the sample image, and determining whether there are any unqualified areas in the area to be inspected in the asphalt concrete core wall based on the local update value of each pixel and the grayscale co-occurrence matrix. This invention fuses local pixel features with global texture features. Compared to the destructive detection of traditional core drilling, this method avoids damage to the asphalt core wall structure, ensures engineering safety, and significantly improves the accuracy of identifying unqualified areas.

Description

Technical Field

[0001] This invention belongs to the field of image processing technology, specifically relating to a method for quality inspection of asphalt concrete core walls based on multimodal fusion. Background Technology

[0002] In the fields of water conservancy and civil engineering, asphalt concrete core walls are a key component of dams and seepage prevention structures. Their quality directly affects the seepage prevention performance and structural safety of the project. Therefore, accurate quality testing is of great significance.

[0003] Traditional methods for testing the quality of asphalt concrete core walls mainly include core sampling and manual visual inspection. Core sampling is a destructive test that not only damages the core wall structure but also has low efficiency and is difficult to achieve large-scale rapid testing. Manual visual inspection relies on the experience of the inspectors, is highly subjective, and is prone to missed detections and misjudgments due to human factors, failing to meet the engineering requirements for testing accuracy and efficiency.

[0004] With the development of computer vision technology, image-based non-destructive inspection methods are gradually being applied to the field of engineering quality inspection. In existing technologies, some methods divide regions by analyzing the gray values ​​of image pixels. However, relying solely on gray value information makes it difficult to distinguish gray value changes caused by factors such as lighting and shooting angle from actual quality defects. This results in limited accuracy in identifying subtle and complex quality defects in asphalt concrete core walls, making it impossible to accurately determine unqualified areas. Summary of the Invention

[0005] To address the above problems, this invention proposes a method for quality inspection of asphalt concrete core walls based on multimodal fusion.

[0006] The technical solution of this invention is: a method for detecting the quality of asphalt concrete core walls based on multimodal fusion, comprising the following steps: S1. Collect sample images of the area to be tested in the asphalt concrete core wall; S2. Determine the local update value of the pixel in the sample image based on the gray value of the pixel and the corresponding neighborhood fitness. S3. Obtain the gray-level co-occurrence matrix of the sample image, and determine whether there are unqualified areas in the area to be inspected of the asphalt concrete core wall based on the local update value of the pixel in the sample image and the gray-level co-occurrence matrix of the sample image.

[0007] Furthermore, S2 includes the following sub-steps: S21. Calculate the neighborhood fitness based on the grayscale change rate of the pixel. S22. Calculate the local update value of the pixel based on the gray value of the pixel and the corresponding neighborhood fitness.

[0008] The beneficial effect of the above-mentioned further solution is that, in this invention, the local pixel features are corrected by combining the pixel grayscale value (reflecting the brightness characteristics of the pixel itself) and the neighborhood fitness (reflecting the grayscale degree of the pixel and the surrounding area).

[0009] Furthermore, S21 includes the following sub-steps: S211. Extract the grayscale change rate of the pixel in the horizontal direction and the grayscale change rate in the vertical direction, and use them as the column vector of the pixel. S212. Use the ratio between the standard deviation of the gray values ​​of all pixels in the D neighborhood of the pixel and the gray value of the pixel as an adjustment factor. S213. Calculate the neighborhood fitness of a pixel based on its column vector and adjustment factor.

[0010] The beneficial effects of the above-mentioned further scheme are as follows: In this invention, the gray-level change rate (gradient) reflects the degree of gray-level abrupt change of a pixel in the horizontal and vertical directions, which is a typical feature of defect edges and heterogeneous regions, and is used as a column vector. The adaptive adjustment factor is used as the weight for fusing directional gray-level changes to determine the degree of adaptation between the pixel and its neighborhood, so that the quantification of neighborhood fitness is more in line with the actual texture characteristics of the asphalt core wall.

[0011] Furthermore, in S212, pixels neighborhood fitness The expression is: ; in, Represents pixels column vectors, Represents pixels Pixels in the local D neighborhood column vectors, This indicates F2 norm processing. This indicates the processing of exponential functions. Represents pixels The regulatory factor.

[0012] Furthermore, S22 includes the following sub-steps: S221. Extract the gray values ​​of all pixels in the 4-neighborhood of the pixel, sort them in descending order as the feature sequence of the pixel, and sort the neighborhood fitness of all pixels according to the descending order as the fitness sequence of the pixel. S222. Generate local update values ​​for pixels based on the feature sequence and fitness sequence of the pixels.

[0013] The beneficial effects of the above-mentioned further solution are: In this invention, the gray values ​​of the 4 neighborhoods are arranged in descending order, which can give priority to highlighting the neighborhood pixels with large gray value differences (these pixels contribute more to defect identification); the neighborhood fitness is arranged accordingly so that the degree of gray value difference and the degree of neighborhood fitness are matched one by one.

[0014] Furthermore, in S222, pixels Local update value The expression is: ; in, Represents pixels grayscale value, Represents pixels neighborhood fitness Represents pixels The first element in the feature sequence, Represents pixels The first element in the fitness sequence, Represents pixels The second element in the feature sequence, Represents pixels The second element in the fitness sequence, Represents pixels The third element in the characteristic sequence, Represents pixels The third element in the fitness sequence, Represents pixels The fourth element in the feature sequence, Represents pixels The fourth element in the fitness sequence.

[0015] The beneficial effect of the above-mentioned further solution is that, in this invention, the ratio of neighborhood fitness is used as a weight to weight the grayscale difference between itself and its neighboring pixels.

[0016] Furthermore, S3 includes the following sub-steps: S31. Obtain the gray-level co-occurrence matrix of the sample image; S32. Extract the trace of the covariance matrix corresponding to the gray-level co-occurrence matrix, and use it as the global gray-level coefficient of the sample image. S33. Determine whether the ratio between the local update value and the gray value of a pixel is greater than or equal to the global gray value coefficient. If so, the pixel is considered an unqualified area in the asphalt concrete core wall inspection area; otherwise, the pixel is considered a qualified area in the asphalt concrete core wall inspection area.

[0017] The beneficial effects of the above-mentioned further solution are as follows: In this invention, the gray-level co-occurrence matrix statistically analyzes the spatial distribution frequency of pixel pairs with different gray values ​​in the image, and then uses the trace of the covariance matrix of the gray-level co-occurrence matrix to reflect the overall dispersion of the gray-level distribution. The ratio of the local update value to the gray value is used to reflect the degree of change of the local pixel. When the degree of local change is greater than or equal to the global dispersion, it indicates that the local anomaly of the pixel region exceeds the normal fluctuation range of the global texture and is judged as an unqualified region.

[0018] Furthermore, in S32, the global grayscale coefficient of the sample image The expression is: ; in, Represents the trace of a matrix. This represents the covariance matrix corresponding to the gray-level co-occurrence matrix. This represents the dimension of the gray-level co-occurrence matrix. This indicates the processing of exponential functions.

[0019] The beneficial effects of this invention are as follows: This invention fuses local pixel features with global texture features, wherein the local pixel features include the grayscale values ​​of pixels, and the global texture features include the grayscale co-occurrence matrix of the core wall sample image; this allows the invention to capture local grayscale abrupt changes in defect areas, breaking through the limitations of traditional single features (only grayscale or only texture) in defect identification. Compared with the destructive detection of traditional core drilling sampling, this invention avoids damage to the asphalt core wall structure, ensures engineering safety, and significantly improves the accuracy of identifying unqualified areas. Attached Figure Description

[0020] Figure 1 This is a flowchart of a method for quality inspection of asphalt concrete core walls based on multimodal fusion. Detailed Implementation

[0021] The embodiments of the present invention will be further described below with reference to the accompanying drawings.

[0022] like Figure 1 As shown, this invention provides a method for quality inspection of asphalt concrete core walls based on multimodal fusion, comprising the following steps: S1. Collect sample images of the area to be tested in the asphalt concrete core wall; S2. Determine the local update value of the pixel in the sample image based on the gray value of the pixel and the corresponding neighborhood fitness. S3. Obtain the gray-level co-occurrence matrix of the sample image, and determine whether there are unqualified areas in the area to be inspected of the asphalt concrete core wall based on the local update value of the pixel in the sample image and the gray-level co-occurrence matrix of the sample image.

[0023] In this embodiment of the invention, S2 includes the following sub-steps: S21. Calculate the neighborhood fitness based on the grayscale change rate of the pixel. S22. Calculate the local update value of the pixel based on the gray value of the pixel and the corresponding neighborhood fitness.

[0024] In this invention, local pixel features are modified by combining pixel grayscale values ​​(reflecting the brightness characteristics of the pixel itself) and neighborhood fitness (reflecting the grayscale degree of the pixel and its surrounding area).

[0025] In this embodiment of the invention, S21 includes the following sub-steps: S211. Extract the grayscale change rate of the pixel in the horizontal direction and the grayscale change rate in the vertical direction, and use them as the column vector of the pixel. S212. Use the ratio between the standard deviation of the gray values ​​of all pixels in the D neighborhood of the pixel and the gray value of the pixel as an adjustment factor. S213. Calculate the neighborhood fitness of a pixel based on its column vector and adjustment factor.

[0026] In this invention, the grayscale change rate (gradient) reflects the degree of grayscale abrupt change of a pixel in the horizontal and vertical directions, which is a typical feature of defect edges and heterogeneous regions, and is used as a column vector. An adaptive adjustment factor is used as the weight for fusing directional grayscale changes to determine the degree of adaptation between a pixel and its neighborhood, so that the quantification of neighborhood fitness better reflects the actual texture characteristics of the asphalt core wall.

[0027] In this embodiment of the invention, in S212, the pixel point neighborhood fitness The expression is: ; in, Represents pixels column vectors, Represents pixels Pixels in the local D neighborhood column vectors, This indicates F2 norm processing. This indicates the processing of exponential functions. Represents pixels The regulatory factor.

[0028] In this embodiment of the invention, S22 includes the following sub-steps: S221. Extract the gray values ​​of all pixels in the 4-neighborhood of the pixel, sort them in descending order as the feature sequence of the pixel, and sort the neighborhood fitness of all pixels according to the descending order as the fitness sequence of the pixel. S222. Generate local update values ​​for pixels based on the feature sequence and fitness sequence of the pixels.

[0029] In this invention, the gray values ​​of the 4-neighborhood are arranged in descending order, which can give priority to highlighting the neighboring pixels with large gray value differences (these pixels contribute more to defect identification); the neighborhood fitness is arranged accordingly so that the degree of neighborhood fitness of gray value difference is matched one by one.

[0030] In this embodiment of the invention, in S222, the pixel point Local update value The expression is: ; in, Represents pixels grayscale value, Represents pixels neighborhood fitness Represents pixels The first element in the feature sequence, Represents pixels The first element in the fitness sequence, Represents pixels The second element in the feature sequence, Represents pixels The second element in the fitness sequence, Represents pixels The third element in the characteristic sequence, Represents pixels The third element in the fitness sequence, Represents pixels The fourth element in the feature sequence, Represents pixels The fourth element in the fitness sequence.

[0031] In this invention, the ratio of neighborhood fitness is used as a weight to weight the grayscale difference between itself and its neighboring pixels.

[0032] In this embodiment of the invention, S3 includes the following sub-steps: S31. Obtain the gray-level co-occurrence matrix of the sample image; S32. Extract the trace of the covariance matrix corresponding to the gray-level co-occurrence matrix, and use it as the global gray-level coefficient of the sample image. S33. Determine whether the ratio between the local update value and the gray value of a pixel is greater than or equal to the global gray value coefficient. If so, the pixel is considered an unqualified area in the asphalt concrete core wall inspection area; otherwise, the pixel is considered a qualified area in the asphalt concrete core wall inspection area.

[0033] In this invention, the gray-level co-occurrence matrix (GLCM) is used to statistically analyze the spatial distribution frequency of pixel pairs with different gray values ​​in an image. The trace of the covariance matrix of the GLCM then reflects the overall dispersion of the gray-level distribution. The ratio of the local update value to the gray value is used to reflect the degree of change in local pixels. When the degree of local change is greater than or equal to the global dispersion, it indicates that the local anomaly in that pixel region exceeds the normal fluctuation range of the global texture and is therefore judged as an unqualified region.

[0034] In this embodiment of the invention, in S32, the global grayscale coefficient of the sample image... The expression is: ; in, Represents the trace of a matrix. This represents the covariance matrix corresponding to the gray-level co-occurrence matrix. This represents the dimension of the gray-level co-occurrence matrix. This indicates the processing of exponential functions.

[0035] Those skilled in the art will recognize that the embodiments described herein are intended to help the reader understand the principles of the invention, and should be understood that the scope of protection of the invention is not limited to such specific statements and embodiments. Those skilled in the art can make various other specific modifications and combinations based on the technical teachings disclosed in this invention without departing from the spirit of the invention, and these modifications and combinations are still within the scope of protection of this invention.

Claims

1. A method for quality inspection of asphalt concrete core walls based on multimodal fusion, characterized in that, Includes the following steps: S1. Collect sample images of the area to be tested in the asphalt concrete core wall; S2. Determine the local update value of the pixel in the sample image based on the gray value of the pixel and the corresponding neighborhood fitness. S3. Obtain the gray-level co-occurrence matrix of the sample image, and determine whether there are unqualified areas in the area to be inspected of the asphalt concrete core wall based on the local update value of the pixel in the sample image and the gray-level co-occurrence matrix of the sample image.

2. The method for quality inspection of asphalt concrete core walls based on multimodal fusion according to claim 1, characterized in that, S2 includes the following sub-steps: S21. Calculate the neighborhood fitness based on the grayscale change rate of the pixel. S22. Calculate the local update value of the pixel based on the gray value of the pixel and the corresponding neighborhood fitness.

3. The method for quality inspection of asphalt concrete core walls based on multimodal fusion according to claim 2, characterized in that, S21 includes the following sub-steps: S211. Extract the grayscale change rate of the pixel in the horizontal direction and the grayscale change rate in the vertical direction, and use them as the column vector of the pixel. S212. Use the ratio between the standard deviation of the gray values ​​of all pixels in the D neighborhood of the pixel and the gray value of the pixel as an adjustment factor. S213. Calculate the neighborhood fitness of a pixel based on its column vector and adjustment factor.

4. The method for detecting the quality of asphalt concrete core walls based on multimodal fusion according to claim 3, characterized in that, In S212, the pixel points neighborhood fitness The expression is: ； in, Represents pixels column vectors, Represents pixels Pixels in the local D neighborhood column vectors, This indicates F2 norm processing. This indicates the processing of exponential functions. Represents pixels The regulatory factor.

5. The method for quality inspection of asphalt concrete core walls based on multimodal fusion according to claim 2, characterized in that, S22 includes the following sub-steps: S221. Extract the gray values ​​of all pixels in the 4-neighborhood of the pixel, sort them in descending order as the feature sequence of the pixel, and sort the neighborhood fitness of all pixels according to the descending order as the fitness sequence of the pixel. S222. Generate local update values ​​for pixels based on the feature sequence and fitness sequence of the pixels.

6. The method for quality inspection of asphalt concrete core walls based on multimodal fusion according to claim 5, characterized in that, In S222, the pixel points Local update value The expression is: ； in, Represents pixels grayscale value, Represents pixels neighborhood fitness Represents pixels The first element in the feature sequence, Represents pixels The first element in the fitness sequence, Represents pixels The second element in the feature sequence, Represents pixels The second element in the fitness sequence, Represents pixels The third element in the characteristic sequence, Represents pixels The third element in the fitness sequence, Represents pixels The fourth element in the feature sequence, Represents pixels The fourth element in the fitness sequence.

7. The method for quality inspection of asphalt concrete core walls based on multimodal fusion according to claim 1, characterized in that, S3 includes the following sub-steps: S31. Obtain the gray-level co-occurrence matrix of the sample image; S32. Extract the trace of the covariance matrix corresponding to the gray-level co-occurrence matrix, and use it as the global gray-level coefficient of the sample image. S33. Determine whether the ratio between the local update value and the gray value of a pixel is greater than or equal to the global gray value coefficient. If so, the pixel is considered an unqualified area in the asphalt concrete core wall inspection area; otherwise, the pixel is considered a qualified area in the asphalt concrete core wall inspection area.

8. The method for quality inspection of asphalt concrete core walls based on multimodal fusion according to claim 7, characterized in that, In step S32, the global grayscale coefficient of the sample image The expression is: ； in, Represents the trace of a matrix. This represents the covariance matrix corresponding to the gray-level co-occurrence matrix. This represents the dimension of the gray-level co-occurrence matrix. This indicates the processing of exponential functions.

Images