A cable shielding layer weaving density detection system based on image analysis

Abstract

The application belongs to the technical field of image processing, and particularly relates to a cable shielding layer weaving density detection system based on image analysis, which comprises an image correction module, a frequency domain analysis module, an adaptive filtering module and a density defect detection module; a maximum peak point of a centralized complex frequency spectrum is obtained; a band-stop filter is constructed according to the maximum peak point, and the centralized complex frequency spectrum is filtered through the band-stop filter to obtain a filtered complex frequency spectrum and perform inverse Fourier transform to obtain a filtered spatial image, and the filtered spatial image is binarized through a threshold segmentation algorithm to obtain a pore mask; a porosity is obtained according to the ratio of non-pore pixel points to the total number of pixel points; and the shielding layer weaving defect is judged according to the porosity. The application automatically identifies and filters the strong periodicity signal of the weaving texture in the frequency domain, effectively highlights the non-periodic defects such as pores, solves the technical problem that the periodic texture covers the defect signal, and improves the accuracy of defect detection.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of image processing. More particularly, the present application relates to a system for detecting the weaving density of a cable shielding layer based on image analysis. BACKGROUND

[0002] In order to resist external electromagnetic interference or prevent internal signal leakage, a cable shielding layer is added to the outside of the cable during production. The shielding performance of the shielding layer is closely related to its weaving density. Maintaining the consistency of the weaving density is crucial to product quality during cable production. Currently, the detection of weaving density mainly relies on manual visual inspection or offline sampling inspection. Manual visual inspection is highly subjective and inefficient, and long-term work can easily lead to fatigue and missed detection. Although offline sampling inspection is accurate, it is a destructive measurement and cannot achieve full inspection. Therefore, there is an urgent need in the industry for a non-contact and automated weaving density detection method.

[0003] In related technologies, machine vision technology is an ideal way to achieve automatic detection. However, there are certain difficulties when applying vision technology to cable shielding layer detection. The cylindrical structure of the cable itself can cause perspective distortion during image acquisition, making the texture inconsistent. In addition, the complex processing texture of the woven wire itself can also produce a large number of non-target details on the image. For example, using Otsu threshold segmentation on the cable shielding layer image cannot find a global threshold that can simultaneously separate the pores and wires, resulting in a fragmented segmentation result that cannot reflect the true pore morphology. This fragmented segmentation result makes it impossible to accurately extract the pore region through morphological analysis. At the same time, the Canny operator is sensitive to noise and strong edges, which not only detects the true edges of the pores, but also detects a large number of edges produced by the processing texture of the woven wire itself. These false edge information causes great interference, making it impossible to effectively analyze the topological structure of the pores, and thus making it difficult to accurately calculate the weaving density. SUMMARY

[0004] To solve the technical problems of the above-mentioned cable periodic weaving texture signal interference defect detection and the existence of motion blur and geometric distortion in collected images, the present application provides a cable shielding layer weaving density detection system based on image analysis, which comprises the following modules: an image correction module for obtaining an input image; a frequency domain analysis module for obtaining a centralized complex spectrum by using Fourier transform on the input image, and obtaining the maximum peak point of the centralized complex spectrum; an adaptive filtering module for constructing a band-stop filter according to the maximum peak point, and filtering the centralized complex spectrum through the band-stop filter to obtain a filtered complex spectrum; the value of any point in the band-stop filter is obtained according to the distance between the point and the maximum peak point, and the distance between the point and the symmetric point of the maximum peak point; a density defect detection module for obtaining a filtered spatial image by performing inverse Fourier transform on the filtered complex spectrum, and obtaining a pore mask by binarizing the filtered spatial image through a threshold segmentation algorithm, wherein the pixel points with a value of 1 in the pore mask are pore pixel points, and the other pixel points are non-pore pixel points; obtaining the porosity according to the ratio of the non-pore pixel points to the total number of pixel points; and judging the weaving defect of the shielding layer according to the porosity.

[0005] The present application automatically searches and locates the main frequency peak point representing the periodic texture of the weaving wire in the centralized complex spectrum through the frequency domain analysis module, solving the technical problem of the related art that the periodic weaving texture signal of the cable shielding layer covers up the weak non-periodic defect signal such as pores and broken wires. The present application further constructs a band-stop filter according to the positions of the identified peak point and its symmetric point through the adaptive filtering module, which ensures that the suppression region of the filter is accurately aligned with the main energy information peak carrying the periodic texture of the weaving wire, and performs filtering processing on the centralized complex spectrum to eliminate these strong interference signals. The non-periodic defect signal such as pores that was originally covered up is retained and highlighted in the filtered complex spectrum. Subsequently, the density defect detection module converts the spectrum graph highlighting the defect information back to a filtered spatial image through inverse Fourier transform. In this spatial image, the defect region such as pores is highlighted, so that the pore mask can be directly binarized through the threshold segmentation algorithm for accurate extraction. Finally, the present application obtains the porosity by calculating the ratio of the non-pore pixel points to the total number of pixel points in the pore mask, and judges the weaving defect according to the porosity, realizing the separation and extraction of the non-periodic defect signal from the strong periodic background, and improving the detection sensitivity and accuracy of the weaving density anomaly and pore defect.

[0006] Preferably, the Fourier transform on the input image to obtain the centralized complex spectrum comprises: applying a Hanning window to the input image, performing fast Fourier transform to obtain an original complex spectrum, and performing centralization processing on the original complex spectrum to obtain the centralized complex spectrum.

[0007] Preferably, the step of obtaining the maximum peak point of the centered complex spectrum includes: obtaining the amplitude spectrum of the centered complex spectrum, performing a logarithmic transformation on the amplitude spectrum, and taking the four largest peak points outside the central low-frequency region in the result of the logarithmic transformation as the maximum peak points.

[0008] This invention addresses the problem that low-amplitude peaks are difficult to detect because the energy in the spectrum is often much higher than in other regions, leading to their difficulty in being detected. The logarithmic transformation effectively compresses the dynamic range of the spectrum, significantly enhancing the visibility of low-amplitude peaks and making them easier to observe and detect. Furthermore, by limiting the search range to the four largest peaks outside the central low-frequency region, this invention ensures that the located peaks truly represent the cross-woven texture in two directions and their conjugate symmetrical peaks, rather than the central zero-frequency component or noise, thus improving the accuracy of peak point localization.

[0009] Preferably, the band-stop filter satisfies the following relationship: In the formula, The first in the band-stop filter The value of each point, The first in the band-stop filter The point and the first Euclidean distance between the maximum peak points The first in the band-stop filter The point and the first Euclidean distance between the points symmetrical to the maximum peak points The number of maximum peak points. For the band-stop filter, regarding the first Bandwidth at the maximum peak point It is an exponential function with the natural constant as the base.

[0010] This invention constructs a band-stop filter by utilizing the distance between a pixel and the maximum peak point, achieving precise filtering of periodic texture frequencies in braided yarns. It effectively utilizes a Gaussian function term to construct the filter's suppression characteristics; the closer a point on the filter is to the maximum peak point or its symmetrical point, the closer the Gaussian function term is to 1, thus making the value at that point smaller and achieving a suppression effect. By multiplying the suppression terms of all peak points and their symmetrical points together, it ensures that all frequency components related to the braided texture are effectively suppressed; simultaneously, this band-stop filter maintains a high pass rate for frequencies far from the peak point, achieving effective separation of periodic and non-periodic signals.

[0011] Preferably, constructing the band-stop filter based on the maximum peak point includes: the size of the band-stop filter is the same as the size of the centered complex spectrum.

[0012] Preferably, the filtering the centered complex spectrum by the band-stop filter comprises: multiplying the band-stop filter with the centered complex spectrum point by point to obtain a filtered complex spectrum; wherein each point in the centered complex spectrum has a complex number value, each value in the band-stop filter is a real number, and the multiplication result of the two numbers is a new real part obtained by multiplying the real number by the real part of the complex number, a new imaginary part obtained by multiplying the real number by the imaginary part of the complex number, and the new real part and the new imaginary part are combined to form a value of a point in the filtered complex spectrum.

[0013] Preferably, the symmetric point of the maximum peak point is a pixel point symmetric to the maximum peak point about a center point of the centered complex spectrum.

[0014] The application solves the problem caused by the conjugate symmetry of Fourier transform by including the pixel point symmetric to the maximum peak point about the center point of the centered complex spectrum in the filtering range; for a real number space image, the amplitude spectrum of its Fourier transform is necessarily symmetric about the origin; therefore, for any one main frequency peak value point representing a woven texture, there is a peak value point with equal energy at its symmetric position; therefore, the application ensures that all energy in the frequency direction is filtered out by simultaneously suppressing the two points, and achieves complete suppression of the woven texture signal.

[0015] Preferably, the distance is the Euclidean distance.

[0016] Preferably, the threshold segmentation algorithm is the Otsu algorithm.

[0017] The application can automatically calculate and determine an optimal global threshold by using the Otsu algorithm to binarize the filtered space image, so as to maximize the separation of two types of pixel points, i.e., pore pixel points and background non-pore pixel points in the image; the application avoids the subjectivity and inadaptability caused by manually setting a fixed threshold by using the adaptive threshold method, and ensures that the pore area highlighted in the filtered space image can be accurately segmented and extracted, thereby providing a reliable pore mask for subsequent statistical porosity.

[0018] Preferably, the judging the woven defect of the shielding layer according to the porosity comprises: when the porosity is greater than a threshold value, the cable shielding layer has a woven defect.

[0019] The beneficial effects of the present application are as follows: the present application effectively overcomes the motion blur problem caused by the cable when collecting at high speed, and the geometric distortion caused by the surface characteristics of the cable cylinder through the image correction module; secondly, the present application converts a detection problem difficult to solve in the spatial domain into a clear and distinct signal separation problem in the frequency domain; at the same time, the present application searches and locates the maximum peak point representing the current braided texture through the frequency domain analysis module, and constructs a band-stop filter based on the peak value information, so that the present application can automatically adapt to cables of different models, different braid angles or with process fluctuations, and has certain detection robustness and industrial applicability; finally, the present application improves the sensitivity and accuracy of detection by threshold segmentation and porosity calculation on the filtered image. BRIEF DESCRIPTION OF DRAWINGS

[0020] Figure 1 is a system block diagram illustrating a cable shielding layer braid density detection system based on image analysis in the present application;

[0021] Figure 2 is an input image in the present application;

[0022] Figure 3 is a detection effect diagram of an existing algorithm;

[0023] Figure 4 is a detection effect diagram of the present application. DETAILED DESCRIPTION

[0024] The technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor fall within the scope of protection of the present application.

[0025] The specific embodiments of the present application will be described in detail below with reference to the drawings.

[0026] The present application provides a cable shielding layer braid density detection system based on image analysis. As shown in Figure 1 A cable shielding layer braid density detection system based on image analysis includes an image correction module 100, a frequency domain analysis module 200, an adaptive filtering module 300 and a density defect detection module 400, which will be described in detail below.

[0027] The image correction module 100 is used to obtain a linear array scanning image of the cable to obtain an input image.

[0028] It should be noted that directly collecting the image of the produced cable will have the problem of motion blur caused by the movement of the cable; at the same time, the geometric distortion caused by the cylindrical surface of the cable makes the braided texture at the central axis of the cable appear wider, and the texture deviating from the central axis appears narrower, which affects the subsequent image analysis process. Therefore, motion compensation and geometric unfolding need to be performed first to convert it into an image with constant period and flat texture.

[0029] Specifically, a line array camera is deployed directly above the cable production line to ensure that the scanning line of the camera is perpendicular to the movement direction of the cable. A high-frequency pulsed light source is used to trigger the camera synchronously to avoid motion blur. As the cable moves, the line array camera collects the shielding layer image row by row to obtain the input image.

[0030] Schematically, Figure 2 is the input image in the present application.

[0031] The frequency domain analysis module 200 performs Fourier transform on the input image to obtain the original complex frequency spectrum and the maximum peak point.

[0032] It should be noted that after processing by the image correction module, the obtained image will exhibit strong periodicity formed by the intersection of two main directions. According to the characteristics of Fourier transform, a specific directional line in the spatial domain will exhibit a peak value on a straight line perpendicular to the line in the frequency domain. Therefore, the periodic braided pattern formed by the intersection of two directions will produce two main high-energy peak points on the frequency spectrum. At the same time, according to the conjugate symmetry property of Fourier transform, the amplitude spectrum of any gray image must be symmetric about the origin. This means that any peak point appearing in the frequency spectrum will necessarily have an equal-energy peak point at its symmetric position. Therefore, the two main peaks representing the two braiding directions will correspond to two additional symmetric peaks, resulting in a total of four high-energy peak points on the frequency spectrum. Therefore, the peak points representing the braided wire texture are found by the frequency domain analysis module to exclude the interference of the braided wire texture on the detection of braided density.

[0033] Specifically, the Hanning window is applied to the input image, and the fast Fourier transform is performed to obtain the original complex frequency spectrum. The original complex frequency spectrum is centered to obtain the centered complex frequency spectrum, and the amplitude spectrum of the centered complex frequency spectrum is obtained. The amplitude spectrum is logarithmically transformed, and the four largest peak points outside the center low-frequency region in the logarithmically transformed result are the maximum peak points.

[0034] It should be further explained that the purpose of applying the Hanning window is to reduce edge effects such as spectral leakage introduced by edge pixels during Fourier transform. Since the energy in the spectrum is usually much higher than in other areas, the logarithmic transform can effectively enhance the visibility of low-amplitude peaks, making them easier to observe and detect. Finally, the centering process rearranges the spectrum of the default Fourier transform output, moving the zero-frequency components from the default arrangement at the four corners of the image to the center of the spectrum, in order to conform to human visual analysis habits and facilitate subsequent peak search and filter design with the center as the origin.

[0035] The adaptive filtering module 300 is used to construct a band-stop filter to eliminate the periodic signal of the braided yarn.

[0036] It should be noted that the peak points in the centered amplitude spectrum carry the main energy information of the periodic texture of the braided wire. In order to reduce the influence of the periodic texture of the braided wire on the braid density detection, this invention constructs a band-stop filter through the peak points in the centered amplitude spectrum to suppress the frequency components represented by these peaks, thereby filtering out the periodic signal of the braided wire, so that non-periodic signals such as the pores of the braided layer can be retained, thereby finding the areas of abnormal braid density in the cable shielding layer.

[0037] Specifically, a band-stop filter with the same size as the centered complex spectrum is constructed. For any point on the band-stop filter, the Euclidean distance between that point and the maximum peak point, as well as the Euclidean distance between that point and the symmetrical point of the maximum peak point about the center position of the centered complex spectrum, are calculated. The value of each point in the band-stop filter is obtained based on the Euclidean distance between the point on the band-stop filter and the maximum peak point, and the distance between the point and the symmetrical point of the maximum peak point.

[0038] Specifically, the band-stop filter satisfies the following relationship:

[0039] ;

[0040] In the formula, The first in the band-stop filter The value of each point, The first in the band-stop filter The point and the first Euclidean distance between the maximum peak points The first in the band-stop filter The point and the first Euclidean distance between the points symmetrical to the maximum peak points The number of maximum peak points. For the band-stop filter, regarding the first Bandwidth at the maximum peak point It is an exponential function with the natural constant as the base.

[0041] in, The range of values ​​is , and These are two Gaussian function terms in the filter, when the... The closer the point is to the first At the point of maximum peak, the closer the Gaussian function term is to 1, the better the band-stop filter becomes. The smaller the value of the first point, the better to filter out the periodic signal of the braided yarn; when the first point... The farther away each point is from the first At the point of maximum peak, the closer the Gaussian function term is to 0, the better the band-stop filter becomes. The larger the value of each point, the better to preserve the non-periodic signal of the pores in the braided layer.

[0042] For example, the bandwidth of the band-stop filter with respect to the maximum peak point is 30, and the implementer can adjust the bandwidth according to the actual situation.

[0043] Furthermore, filtering the centered complex spectrum using a band-stop filter includes: multiplying the band-stop filter by the centered complex spectrum point by point to obtain the filtered complex spectrum; wherein, the value of each point in the centered complex spectrum is a complex number, and the value of each point in the band-stop filter is a real number. The result of multiplying the two numbers is a real number multiplied by the real part of the complex number to obtain a new real part, and a real number multiplied by the imaginary part of the complex number to obtain a new imaginary part. The new real part and the new imaginary part are combined to obtain the value of a point in the filtered complex spectrum.

[0044] The density defect detection module 400 is used to count the pore area and calculate the weaving density.

[0045] It should be noted that only aperiodic components, such as pores, remain in the filtered complex spectrum. Performing an inverse Fourier transform on this spectrum, the periodic woven texture in the resulting spatial image is greatly suppressed, while defects such as pores and broken fibers are preserved and highlighted as aperiodic signals. At this point, the pore region will appear as high-energy spots. By performing simple thresholding on the obtained spatial image, the pore mask can be accurately extracted, and the pore area ratio can be statistically analyzed.

[0046] Specifically, an inverse Fourier transform is performed on the filtered complex spectrum to obtain the filtered spatial image. The filtered spatial image is then binarized using the Otsu thresholding algorithm to obtain the aperture mask. Pixels with a value of 1 in the aperture mask are aperture pixels, and the other pixels are non-aperture pixels.

[0047] Furthermore, the porosity is defined as the ratio between the number of non-porous pixels and the total number of pixels in the image. When the porosity is greater than a threshold of 0.1, the cable shielding layer has a braiding defect. The implementation personnel can adjust the threshold according to the actual situation.

[0048] Exemplarily, Figure 3 The detection effect diagram of the prior art algorithm can be seen from the image that the prior art algorithm can only segment the macroscopic contour of the braided wire itself, and cannot effectively separate the pore defect from the normal texture background, Figure 4 The detection effect diagram of the present application, in the detection effect diagram of the present application, the strong periodic texture signal of the braided wire is suppressed, and the non-periodic signal such as the pore is retained and highlighted.

[0049] The above are preferred embodiments of the present application, and do not limit the protection scope of the present application, so: any equivalent changes made according to the structure and principle of the present application should be covered within the protection scope of the present application.

Claims

1. An image analysis based cable shield braid density detection system, characterized by, The application relates to a method for detecting the weaving defects of a shielding layer of a cable, comprising the following steps: An image correction module is used to acquire an input image; A frequency domain analysis module is used to acquire a centralized complex frequency spectrum by using Fourier transform on the input image, and to acquire maximum peak points of the centralized complex frequency spectrum, the steps comprising: acquiring an amplitude spectrum of the centralized complex frequency spectrum, and performing logarithmic transformation on the amplitude spectrum to obtain the maximum peak points as four maximum peak points outside a central low-frequency area in the logarithmic transformation result; An adaptive filtering module is used to construct a band-stop filter according to the maximum peak points, and to filter the centralized complex frequency spectrum by using the band-stop filter to obtain a filtered complex frequency spectrum; the value of any point in the band-stop filter is acquired according to the distance between the point and the maximum peak points and the distance between the point and the symmetric points of the maximum peak points; A Hanning window is applied to reduce the edge effect of spectrum leakage caused by the edge pixels of the image when Fourier transform is performed; The band-stop filter satisfies the relationship: ; in which, is the value of the first point in the band-stop filter, is the Euclidean distance between the first point and the first maximum peak point in the band-stop filter, is the Euclidean distance between the first point and the symmetric point of the first maximum peak point in the band-stop filter, is the number of maximum peak points, is the bandwidth of the band-stop filter with respect to the first maximum peak point, is an exponential function with a natural constant as the base number; A density defect detection module is used to perform inverse Fourier transform on the filtered complex frequency spectrum to obtain a filtered spatial image, to binarize the filtered spatial image by using a threshold segmentation algorithm to obtain a pore mask, the pixel points with a value of 1 in the pore mask are pore pixel points, and the other pixel points are non-pore pixel points; the porosity is acquired according to the ratio of the non-pore pixel points to the total number of pixel points; and the weaving defects of the shielding layer are judged according to the porosity.

2. The system for detecting the weaving density of the shielding layer of the cable based on image analysis according to claim 1, characterized in that, The Fourier transform on the input image to acquire the centralized complex frequency spectrum comprises: applying the Hanning window to the input image, performing fast Fourier transform to obtain an original complex frequency spectrum, and performing centralization processing on the original complex frequency spectrum to obtain the centralized complex frequency spectrum.

3. The system for detecting the weaving density of the shielding layer of the cable based on image analysis according to claim 1, characterized in that, The band-stop filter is constructed according to the maximum peak points, and the size of the band-stop filter is the same as that of the centralized complex frequency spectrum.

4. The system for detecting the weaving density of the shielding layer of the cable based on image analysis according to claim 1, characterized in that, The filtering of the centralized complex frequency spectrum by using the band-stop filter comprises: multiplying the band-stop filter and the centralized complex frequency spectrum point by point to obtain the filtered complex frequency spectrum; wherein the value of each point in the centralized complex frequency spectrum is a complex number, the value of each point in the band-stop filter is a real number, the multiplication result of the two numbers is a new real part obtained by multiplying a real number by the real part of a complex number, a new imaginary part obtained by multiplying the real number by the imaginary part of the complex number, and the new real part and the new imaginary part are combined to form the value of a point in the filtered complex frequency spectrum.

5. The system for detecting the weaving density of the shielding layer of the cable based on image analysis according to claim 1, characterized in that, The symmetric points of the maximum peak points are pixel points symmetric to the maximum peak points about the center point of the centralized complex frequency spectrum.

6. The image analysis based system for detecting the braid density of a cable shield layer according to claim 1, wherein, The distance is the Euclidean distance.

7. The system for detecting the weaving density of the shielding layer of the cable based on image analysis according to claim 1, characterized in that, The threshold segmentation algorithm is the Otsu algorithm.

8. The image analysis based system for detecting the braid density of a cable shield layer according to claim 1, wherein, When the porosity is greater than a threshold value, the cable shielding layer has weaving defects.

Images