A method for detecting quality of nonwoven fabrics

Abstract

The present invention discloses a non-woven fabric quality inspection method, which belongs to the field of quality inspection technology, and includes the following steps: S1, collecting the production image of the non-woven fabric, and using the homomorphic filtering method to complete preprocessing of the production image to obtain a stable production image; S2, using the grayscale co-occurrence matrix to extract the characteristic primitives of the remaining pixel points in the stable production image except the first row and the last row, and generate texture characteristics; S3, taking the pixel points whose grayscale value is less than the texture characteristic as the quality defect position of the non-woven fabric. The present invention can accurately identify the flaws and defects on the surface of the non-woven fabric; it can save labor costs for enterprises and strictly control the quality of non-woven fabrics.

Description

Technical Field

[0001] The invention belongs to the technical field of quality detection, and in particular relates to a non-woven fabric quality detection method. Background Art

[0002] As a new type of environmentally friendly fabric, non-woven fabric has the advantages of good air permeability, moisture resistance, softness, lightness, non-combustibility, easy decomposition, non-toxicity, non-irritation, recyclability and low price. It is a widely used cloth material. In order to ensure the quality of non-woven fabrics, it is extremely important to conduct defect detection on the surface of non-woven fabrics in the actual production process and remove defective products.

[0003] At present, most factories still use manual inspection to check whether there are defects on the surface of non-woven fabrics. A worker needs to inspect the non-woven fabric surface with a width of more than 4m for 8 hours in insufficient light. However, most non-woven fabric defects are consistent with the color of the fabric surface, which makes it difficult for the human eye to distinguish them. In addition, the speed of naked eye inspection is also very low. Therefore, manual inspection of fabrics is low in accuracy and slow in speed, and it is easy to miss defects and cause defective products to flow out. Summary of the invention

[0004] In order to solve the above problems, the present invention proposes a non-woven fabric quality detection method.

[0005] The technical solution of the present invention is: a non-woven fabric quality detection method comprises the following steps:

[0006] S1. Collect production images of non-woven fabrics, and use homomorphic filtering to pre-process the production images to obtain stable production images;

[0007] S2, using the gray level co-occurrence matrix to extract the characteristic primitives of the remaining pixels except the first row and the last row in the stable production image, and generate texture characteristics;

[0008] S3. Pixels whose grayscale values ​​are smaller than the texture characteristics are regarded as quality defect locations of the non-woven fabric.

[0009] In S1, homomorphic filtering can increase contrast and standardize brightness at the same time by removing multiplicative noise, thereby achieving the purpose of image enhancement.

[0010] Furthermore, S2 includes the following sub-steps:

[0011] S21, extracting the energy value corresponding to the gray-level co-occurrence matrix of the stable production image;

[0012] S22, determining a first difference position and a second difference position in a first row of the stable production image according to the energy value;

[0013] S23, determining a third difference position and a fourth difference position in the last row of the stable production image according to the first difference position and the second difference position in the first row;

[0014] S24, determining characteristic primitives of each pixel point in the second row to the K-1th row in the stable production image according to the first difference position, the second difference position, the third difference position and the fourth difference position, where K represents the number of rows of the stable production image;

[0015] S25. Calculate texture characteristics according to characteristic primitives of each pixel point in the second row to the K-1th row in the stable production image.

[0016] The beneficial effect of the above further scheme is: in the present invention, the energy value refers to the sum of the squares of the element values ​​of the grayscale co-occurrence matrix. By extracting the energy value through the grayscale co-occurrence matrix, the uniformity and smoothness of the texture in the image can be accurately quantified, providing an accurate basis for the subsequent determination of the difference position. By determining the difference position in the first row according to the energy value, the area where the change or abnormality occurs in the image can be accurately located. By determining the corresponding difference position in the last row through the difference position of the first row, it is helpful to maintain the consistency of the image in the vertical direction and improve the accuracy of the analysis. The entire process from grayscale co-occurrence matrix extraction to difference position determination, and then to characteristic primitives and texture characteristic calculation provides a complete framework for the comprehensive analysis of the image, which can fully understand the distribution and changes of texture characteristics in the image.

[0017] Further, in S22, the first difference position is determined by a first difference criterion, and the expression of the first difference criterion Z1 is: ; Where Z1(u) represents the operation of determining the first difference position, u represents the first row of pixels in the stable production image, and C u represents the gray value of the first row of pixel u in the stable production image, E represents the energy value corresponding to the gray-level co-occurrence matrix, H represents the number of columns of the stable production image, and max(·) represents the maximum value operation;

[0018] In S22, the second difference position is determined by a second difference criterion, and the expression of the second difference criterion Z2 is: ; Wherein, Z2(u) represents the operation of determining the second difference position, and min(·) represents the minimum value operation.

[0019] Further, in S23, the uniform contrast between each pixel point in the last row of the stable production image and the first difference position and the second difference is calculated, and the pixel point with the maximum uniform contrast is taken as the third difference position;

[0020] The calculation formula for uniform contrast D is: ; In the formula, e represents the entropy of the gray-level co-occurrence matrix, represents the gray value of the first difference position, represents the gray value of the second difference position, and c represents the gray value of the pixel in the last row of the stable production image;

[0021] In S23, a pixel point that is separated from the third difference position by a distance a is used as a fourth difference position, where a represents a pixel interval of a gray level co-occurrence matrix.

[0022] Pixel spacing refers to the relative distance between two pixels considered when calculating the gray-level co-occurrence matrix. This distance can be in the horizontal, vertical or diagonal direction. The choice of pixel spacing has a significant impact on the calculation results of the gray-level co-occurrence matrix. Pixel spacing usually takes an integer value.

[0023] Further, S24 includes the following sub-steps:

[0024] S241, generating a difference angle for each pixel point in the second row to the K-1th row in the stable production image according to the first difference position, the second difference position, the third difference position and the fourth difference position;

[0025] S242: Generate characteristic primitives according to the difference angles of each pixel point and the direction of the gray-level co-occurrence matrix.

[0026] The beneficial effect of the above further scheme is that: in the present invention, the generation of the difference angle helps to further analyze the changing trend and pattern of the texture in the image, and provides an accurate basis for the subsequent generation of characteristic primitives. The generation of the difference angle not only takes into account the changes in a single pixel, but also combines the information of multiple key positions (i.e., four difference positions) in the image, thereby providing a more comprehensive perspective to analyze the image. Direction θ refers to the relative direction between the two pixels considered when calculating the grayscale co-occurrence matrix. This direction can be horizontal (0°), vertical (90°), 45° or 135°, etc. The characteristic primitive can accurately reflect the local characteristics and global patterns of the texture in the image.

[0027] Further, in S241, if the pixel point belongs to the second row to the If there are rows, the pixel points are connected to the first difference position and the second difference position respectively to generate a difference angle, where K represents the number of rows of the stable production image. Indicates rounding up; if the pixel belongs to From the K-1th row to the K-1th row, the pixel point is connected to the third difference position and the fourth difference position respectively to generate a difference angle.

[0028] Further, in S242, the expression of the characteristic primitive x of the pixel points in the second row to the K-1th row is: ; In the formula, θ represents the direction of the gray-level co-occurrence matrix, β represents the difference angle of the pixel point, Represents the grayscale value of the pixels in the second to K-1th rows, represents the gray value of the first difference position, represents the gray value of the second difference position, represents the gray value of the first difference position, Represents the grayscale value of the second difference position.

[0029] Furthermore, in S25, the calculation formula of the texture characteristic T is: Where, C ave_1 represents the mean gray value of all pixels in the first row of the stable production image, C ave_2 represents the mean gray value of all pixels in the last row of the stable production image, X g It represents the modulus of the characteristic primitives of the remaining pixels in the stable production image except the first and last rows, and G represents the total number of pixels in the stable production image except the first and last rows.

[0030] The characteristic primitives are row vectors whose magnitude is the square root of the sum of the squares of its components.

[0031] The beneficial effects of the present invention are as follows: the present invention utilizes the homomorphic filtering method to process the production images of non-woven fabrics, which can effectively remove noise and interference in the images, while enhancing the contrast and details of the images, making subsequent processing more accurate and stable; the characteristic primitives extracted by the grayscale co-occurrence matrix can comprehensively reflect the texture characteristics of the non-woven fabric production images, provide a reliable basis for subsequent quality defect detection, and can accurately identify the flaws and defects on the surface of the non-woven fabrics; at the same time, it can save labor costs for enterprises and strictly control the quality of non-woven fabrics. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] Figure 1 Flow chart of nonwoven fabric quality testing method. DETAILED DESCRIPTION

[0033] The embodiments of the present invention will be further described below in conjunction with the accompanying drawings.

[0034] like Figure 1 As shown, the present invention provides a non-woven fabric quality detection method, comprising the following steps:

[0035] S1. Collect production images of non-woven fabrics, and use homomorphic filtering to pre-process the production images to obtain stable production images;

[0036] S2, using the gray level co-occurrence matrix to extract the characteristic primitives of the remaining pixels except the first row and the last row in the stable production image, and generate texture characteristics;

[0037] S3. Pixels whose grayscale values ​​are smaller than the texture characteristics are regarded as quality defect locations of the non-woven fabric.

[0038] In S1, homomorphic filtering can increase contrast and standardize brightness at the same time by removing multiplicative noise, thereby achieving the purpose of image enhancement.

[0039] In this embodiment of the present invention, S2 includes the following sub-steps:

[0040] S21, extracting the energy value corresponding to the gray-level co-occurrence matrix of the stable production image;

[0041] S22, determining a first difference position and a second difference position in a first row of the stable production image according to the energy value;

[0042] S23, determining a third difference position and a fourth difference position in the last row of the stable production image according to the first difference position and the second difference position in the first row;

[0043] S24, determining characteristic primitives of each pixel point in the second row to the K-1th row in the stable production image according to the first difference position, the second difference position, the third difference position and the fourth difference position, where K represents the number of rows of the stable production image;

[0044] S25. Calculate texture characteristics according to characteristic primitives of each pixel point in the second row to the K-1th row in the stable production image.

[0045] In the present invention, the energy value refers to the sum of the squares of the element values ​​of the grayscale co-occurrence matrix. By extracting the energy value through the grayscale co-occurrence matrix, the uniformity and smoothness of the texture in the image can be accurately quantified, providing an accurate basis for the subsequent determination of the difference position. By determining the difference position in the first row according to the energy value, the area where the change or abnormality occurs in the image can be accurately located. By determining the corresponding difference position in the last row through the difference position in the first row, it is helpful to maintain the consistency of the image in the vertical direction and improve the accuracy of the analysis. The entire process from grayscale co-occurrence matrix extraction to difference position determination, and then to characteristic primitives and texture characteristic calculation provides a complete framework for the comprehensive analysis of the image, which can fully understand the distribution and changes of texture characteristics in the image.

[0046] In the embodiment of the present invention, in S22, the first difference position is determined by a first difference criterion, and the expression of the first difference criterion Z1 is: ; Where Z1(u) represents the operation of determining the first difference position, u represents the first row of pixels in the stable production image, and C u represents the gray value of the first row of pixel u in the stable production image, E represents the energy value corresponding to the gray-level co-occurrence matrix, H represents the number of columns of the stable production image, and max(·) represents the maximum value operation;

[0047] In S22, the second difference position is determined by a second difference criterion, and the expression of the second difference criterion Z2 is: ; Wherein, Z2(u) represents the operation of determining the second difference position, and min(·) represents the minimum value operation.

[0048] In the embodiment of the present invention, in S23, the uniform contrast between each pixel point in the last row of the stable production image and the first difference position and the second difference is calculated, and the pixel point with the maximum uniform contrast is used as the third difference position;

[0049] The calculation formula for uniform contrast D is: ; In the formula, e represents the entropy of the gray-level co-occurrence matrix, represents the gray value of the first difference position, represents the gray value of the second difference position, and c represents the gray value of the pixel in the last row of the stable production image;

[0050] In S23, a pixel point that is separated from the third difference position by a distance a is used as a fourth difference position, where a represents a pixel interval of a gray level co-occurrence matrix.

[0051] Pixel spacing refers to the relative distance between two pixels considered when calculating the gray-level co-occurrence matrix. This distance can be in the horizontal, vertical or diagonal direction. The choice of pixel spacing has a significant impact on the calculation results of the gray-level co-occurrence matrix. Pixel spacing usually takes an integer value.

[0052] In this embodiment of the present invention, S24 includes the following sub-steps:

[0053] S241, generating a difference angle for each pixel point in the second row to the K-1th row in the stable production image according to the first difference position, the second difference position, the third difference position and the fourth difference position;

[0054] S242: Generate characteristic primitives according to the difference angles of each pixel point and the direction of the gray-level co-occurrence matrix.

[0055] In the present invention, the generation of difference angles helps to further analyze the changing trends and patterns of textures in images, and provides an accurate basis for the subsequent generation of characteristic primitives. The generation of difference angles not only takes into account the changes in a single pixel, but also combines the information of multiple key positions (i.e., four difference positions) in the image, thereby providing a more comprehensive perspective to analyze the image. Direction θ refers to the relative direction between the two pixels considered when calculating the grayscale co-occurrence matrix. This direction can be horizontal (0°), vertical (90°), 45°, or 135°, etc. The characteristic primitives can accurately reflect the local features and global patterns of textures in images.

[0056] In the embodiment of the present invention, in S241, if the pixel point belongs to the second row to the first row, If there are rows, the pixel points are connected to the first difference position and the second difference position respectively to generate a difference angle, where K represents the number of rows of the stable production image. Indicates rounding up; if the pixel belongs to From the K-1th row to the K-1th row, the pixel point is connected to the third difference position and the fourth difference position respectively to generate a difference angle.

[0057] In the embodiment of the present invention, in S242, the expression of the characteristic primitive x of the pixel points in the second row to the K-1th row is: ; In the formula, θ represents the direction of the gray-level co-occurrence matrix, β represents the difference angle of the pixel point, Represents the grayscale value of the pixels in the second to K-1th rows, represents the gray value of the first difference position, represents the gray value of the second difference position, represents the gray value of the first difference position, Represents the grayscale value of the second difference position.

[0058] In the embodiment of the present invention, in S25, the calculation formula of the texture characteristic T is: Where, C ave_1 represents the mean gray value of all pixels in the first row of the stable production image, C ave_2 represents the mean gray value of all pixels in the last row of the stable production image, X g It represents the modulus of the characteristic primitives of the remaining pixels in the stable production image except the first and last rows, and G represents the total number of pixels in the stable production image except the first and last rows.

[0059] The characteristic primitives are row vectors whose magnitude is the square root of the sum of the squares of its components.

[0060] Those skilled in the art will appreciate that the embodiments described herein are intended to help readers understand the principles of the present invention, and should be understood that the protection scope of the present invention is not limited to such specific statements and embodiments. Those skilled in the art can make various other specific variations and combinations that do not deviate from the essence of the present invention based on the technical revelations disclosed by the present invention, and these variations and combinations are still within the protection scope of the present invention.

Claims

1. A nonwoven fabric quality detection method, characterized in that: The following steps are involved: S1. Collect production images of non-woven fabrics, and use homomorphic filtering to pre-process the production images to obtain stable production images; S2, using the gray level co-occurrence matrix to extract the characteristic primitives of the remaining pixels except the first row and the last row in the stable production image, and generate texture characteristics; S3, taking the pixel points whose gray value is less than the texture characteristic as the quality defect position of the non-woven fabric; The S2 comprises the following sub-steps: S21, extracting the energy value corresponding to the gray-level co-occurrence matrix of the stable production image; S22, determining a first difference position and a second difference position in a first row of the stable production image according to the energy value; S23, determining a third difference position and a fourth difference position in the last row of the stable production image according to the first difference position and the second difference position in the first row; S24, determining characteristic primitives of each pixel point in the second row to the K-1th row in the stable production image according to the first difference position, the second difference position, the third difference position and the fourth difference position, where K represents the number of rows of the stable production image; S25, calculating texture characteristics according to characteristic primitives of each pixel point in the second row to the K-1th row in the stable production image; In S22, the first difference position is determined by a first difference criterion, and the expression of the first difference criterion Z1 is: ; Where Z1(u) represents the operation of determining the first difference position, u represents the first row of pixels in the stable production image, and C u represents the gray value of the first row of pixel u in the stable production image, E represents the energy value corresponding to the gray-level co-occurrence matrix, H represents the number of columns of the stable production image, and max(·) represents the maximum value operation; In S22, the second difference position is determined by a second difference criterion, and the expression of the second difference criterion Z2 is: ; Wherein, Z2(u) represents the operation of determining the second difference position, and min(·) represents the minimum value operation; In S23, the uniform contrast between each pixel point in the last row of the stable production image and the first difference position and the second difference is calculated, and the pixel point with the maximum uniform contrast is used as the third difference position; The calculation formula of the uniform contrast D is: ; In the formula, e represents the entropy of the gray-level co-occurrence matrix, represents the gray value of the first difference position, represents the gray value of the second difference position, and c represents the gray value of the pixel in the last row of the stable production image; In the step S23, a pixel point that is separated from the third difference position by a distance a is used as a fourth difference position, where a represents a pixel interval of a gray level co-occurrence matrix; The S24 comprises the following sub-steps: S241, generating a difference angle for each pixel point in the second row to the K-1th row in the stable production image according to the first difference position, the second difference position, the third difference position and the fourth difference position; S242, generating characteristic primitives according to the difference angles of each pixel point and the direction of the gray-level co-occurrence matrix; In S241, if the pixel point belongs to the second row to the first row, If there are rows, the pixel points are connected to the first difference position and the second difference position respectively to generate a difference angle, where K represents the number of rows of the stable production image. Indicates rounding up; if the pixel belongs to When the pixel point reaches the K-1th row, the pixel point is connected to the third difference position and the fourth difference position respectively to generate a difference angle; In S242, the expression of the characteristic primitive x of the pixels in the second row to the K-1th row is: ; In the formula, θ represents the direction of the gray-level co-occurrence matrix, β represents the difference angle of the pixel point, Represents the grayscale value of the pixels in the second to K-1th rows, represents the gray value of the first difference position, represents the gray value of the second difference position, represents the gray value of the first difference position, Represents the grayscale value of the second difference position.

2. The nonwoven fabric quality detection method according to claim 1, characterized in that: In S25, the calculation formula of the texture characteristic T is: Where, C ave_1 represents the mean gray value of all pixels in the first row of the stable production image, C ave_2 represents the mean gray value of all pixels in the last row of the stable production image, X g It represents the modulus of the characteristic primitives of the remaining pixels in the stable production image except the first and last rows, and G represents the total number of pixels in the stable production image except the first and last rows.

Images