An improved ecological antibacterial yarn quality monitoring method based on machine vision

Abstract

The application relates to the technical field of antibacterial yarn production, in particular to an improved ecological antibacterial yarn quality monitoring method based on machine vision, which comprises the following steps: deploying a conduction component at the output end of a yarn production device, pushing the yarn output by the output end of the yarn production device based on the conduction component, and collecting yarn image data in the process of yarn transmission in the conduction component; setting a yarn image data acquisition time domain, and continuously acquiring the collected yarn image data based on the acquisition time domain; the application provides a stable collection environment for yarn image data collection by setting a specific conduction component, and brings global similarity analysis to the yarn produced by the yarn production device through specified and continuous yarn image collection logic, so that the production yarn quality balance is evaluated based on the similarity analysis result; compared with the prior art of applying image recognition technology to yarn quality evaluation, the application has higher precision and real-time performance.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of antibacterial yarn production, in particular to an improved ecological antibacterial yarn quality monitoring method based on machine vision. BACKGROUND

[0002] Ecological antibacterial yarn is an innovative textile material based on natural or degradable raw materials and made through special processes. It not only has good skin-friendliness and environmental friendliness, but also can effectively inhibit the growth of various bacteria such as Escherichia coli and Staphylococcus aureus. It is commonly used to make underwear, baby clothes and other products to protect the health of the skin, and it also meets the concept of green living. The invention patent application with application number 202310694846.9 discloses a fiber tape online quality monitoring method based on visual monitoring, which includes the following steps: step P1, image shooting, continuously shooting the fiber tape in the conveying process through a camera to obtain the image of the fiber tape and upload it to an image processing system; step P2, image processing, processing the photographed picture and obtaining the black and white image of the denoised fiber tape; step P3, comparative analysis, comparing the image obtained in step P2 with the pre-set standard image; step P4, judging whether the quality of the fiber tape is qualified; step P5, issuing a preset instruction: the defects that can be considered in the process of image processing include three cases, namely, hole defects, wrinkle defects and crack defects.

[0003] The application aims to solve the problem of "holes, wrinkles and cracks on the surface of the fiber tape during the laying process, which affects the automatic laying quality".

[0004] However, for ecological antibacterial yarn, in order to ensure its performance, the quality of its production process needs to be monitored. The prior art has examples of sampling production yarn images and applying image recognition technology to monitor the quality of yarn production, but it has defects: the sampled images are too limited in their relevance, making it difficult to reflect the overall quality of the production yarn.

[0005] Therefore, an improved ecological antibacterial yarn quality monitoring method based on machine vision is proposed. SUMMARY

[0006] In view of the above-mentioned shortcomings of the prior art, the present application provides an improved ecological antibacterial yarn quality monitoring method based on machine vision, which solves the technical problems raised in the background art.

[0007] To achieve the above purpose, the present application is realized by the following technical scheme:

[0008] An improved ecological antibacterial yarn quality monitoring method based on machine vision, comprising:

[0009] A conducting assembly is arranged at an output end of a yarn production device, yarn outputted from the output end of the yarn production device is pushed based on the conducting assembly, yarn image data is collected in the process of the yarn being transmitted in the conducting assembly; a yarn image data acquisition time domain is set, the collected yarn image data is continuously acquired based on the acquisition time domain, yarn quality assessment is performed once when the time stamp corresponding to the earliest acquired yarn image data and the currently acquired yarn image data matches the set yarn image data acquisition time domain; yarn image data is acquired, yarn images are segmented from each yarn image data, and the similarity of each yarn image is compared;

[0010] The similarity comparison result of each yarn image is acquired, the production quality is assessed based on the comparison result, a yarn qualified judgment threshold is set, and the set yarn qualified judgment threshold and the yarn production quality assessment result are compared to determine whether the currently produced yarn is qualified. Further, the conducting assembly is integrated by a sleeve and a roller, the roller is symmetrically arranged inside the sleeve, the yarn outputted from the output end of the yarn production device enters the sleeve and contacts the surface of the roller, and the yarn is transmitted through the sleeve based on the traction force of the output end of the yarn production device;

[0011] The color of the inner wall of the sleeve and the surface of the roller is a solid color of an RGB color mode different from the color of the yarn, the collection device for collecting yarn image data is arranged inside the sleeve, and the collection device applies an exposure mode to perform collection of yarn image data each time the collection device collects yarn image data inside the sleeve;

[0012] The yarn image data is continuously collected based on a specified frequency, the quality of the collected yarn image data is recognized synchronously after the yarn image data is collected each time, and a yarn image data collection time node is inserted in the yarn image data collection frequency based on the quality of the yarn image data.

[0013] Further, the quality recognition logic of the yarn image data and the decision logic of the inserted yarn image data collection time node are represented as:

[0014]

[0015] In the formula: Q is the quality performance value of the yarn image data; C and E are the clarity index and the exposure accuracy index; ω1 and ω2 are the weights; Q norr is a quality qualified judgment threshold defined based on the quality level of the image data;

[0016] Wherein, the weights ω1, ω2 are positive numbers, and the sum is 1, the weight value is defined by the user end, the weight ω1, ω2 is initially set to 0.6, 0.4, when formula (2) is established, the yarn image data is collected according to the specified frequency, when formula (2) is not established, the latest collected yarn image data collection time stamp is obtained, and the next yarn image data collection time stamp is determined based on the collection frequency, which is denoted as a1, a2, then the inserted yarn image data collection time node is Further, the calculation logic of C, E is represented as:

[0017]

[0018] In the formula: n is the total amount of pixels in the yarn image data; G x , G y is the gradient of the i-th pixel in the x and y directions calculated by the Sobel operator; is the average gray value of the yarn image converted to a gray image; G ideal is the standard gray value;

[0019] Wherein, the standard gray value G ideal is defined by the user end of the system.

[0020] Further, after the yarn image data based on the yarn image data acquisition time domain acquisition operation ends, all the acquired yarn image data is further traversed, and all the yarn image data is sorted based on its acquisition time sequence. The yarn image data collected by inserting the yarn image data collection time node is identified synchronously based on the acquisition time sequence sorting. Further, in the yarn image similarity comparison stage, based on the identification result, the yarn image data collected according to the specified frequency is taken as image data group one, the yarn image data collected according to the inserted yarn image data collection time node is taken as image data group two, the adjacent yarn image data collected according to the specified frequency and the yarn image data collected according to the inserted yarn image data collection time node are taken as image data group, the image data group is composed of several groups, and the similarity comparison is performed on the image data group and the image data group.

[0021] Further, the similarity calculation formula of the image data group is:

[0022]

[0023] In the formula: M, N are the width and height of the yarn image data; F A (u,v), F B (u,v) is the representation result of A and B in the frequency domain after two-dimensional discrete Fourier transform is performed on the image A and B respectively; is F Bthe complex conjugate of (u, v); (u, v) is the frequency domain coordinate;

[0024] wherein all the yarn image data in the image data group one and the image data group two are calculated based on the above formula respectively, and the results are averaged, and are respectively denoted as The similarity calculation formula of the image data group is:

[0025]

[0026] In the formula: c(x, y) is used to determine whether the edge directions of two images in the image data group at the corresponding pixel position (x, y) are consistent; e1(x, y), e2(x, y) are the edge intensity values of the pixel (x, y) in the two edge images obtained by using the Canny edge detection algorithm to perform edge detection on the two images in the image data group; f(·) is a determination function:

[0027] wherein when the pixel (x, y) is an edge pixel, the edge intensity value thereof is 1, otherwise, the edge intensity value thereof is 0, e1(x, y)+e2(x, y)>0, then f(·)=1, otherwise, f(·)=0, and the similarity calculation results in the image data groups are denoted as SIMM'1, SIMM'2, SIMM'3,....

[0028] Further, the value of c(x, y) is subject to:

[0029]

[0030] In the formula: θ1(x, y), θ2(x, y) are the quantized gradient directions of the pixel (x, y) in the two images in the image data group after edge detection; b is the number of direction intervals when the gradient direction is quantized to the direction interval;

[0031] wherein b is initially set as 8.

[0032] Further, when the yarn image is segmented in the yarn image data, the gray value interval of each pixel gray value in the yarn image data is identified through the image background gray value interval and the yarn image gray value interval defined by the user terminal, based on the identification result, all the pixels in the yarn image gray value interval are obtained, and the combination of the obtained pixels is denoted as the segmented yarn image in the yarn image data.

[0033] Further, the yarn production quality evaluation logic is represented as:

[0034]

[0035] In the formula: P is the yarn production quality performance value; α, ε j are weights;

[0036] wherein the weight α, ε j are both greater than zero, and Compared with the known prior art, the technical scheme provided by the present application has the following beneficial effects:

[0037] The present application provides an improved ecological antibacterial yarn quality monitoring method based on machine vision. In the execution process, the method provides a stable collection environment for yarn image data collection by setting a specific transmission component, and performs global similarity analysis on the yarn produced by the yarn production equipment through specified and continuous yarn image collection logic, thereby evaluating the production yarn quality balance based on the similarity analysis result. Compared with the prior art of applying image recognition technology to yarn quality evaluation, the present application has higher precision and real-time performance. BRIEF DESCRIPTION OF DRAWINGS

[0038] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the following will briefly introduce the drawings needed to be used in the embodiment or prior art description. Obviously, the drawings in the following description are only some embodiments of the present application, and for those skilled in the art, other drawings can also be obtained without creative labor on the basis of these drawings.

[0039] Figure 1 A flowchart of an improved ecological antibacterial yarn quality monitoring method based on machine vision;

[0040] Figure 2 An installation diagram of a data collection device:

[0041] The figure marks represent: 1, sleeve; 2, roller; 3, yarn; 4, yarn image data collection device. DETAILED DESCRIPTION

[0042] In order to make the purpose, technical scheme and advantages of the embodiments of the present application more clear, the technical scheme in the embodiments of the present application will be described clearly and completely in combination with the drawings in the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, not all. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor are within the scope of protection of the present application.

[0043] The present application will be further described below in combination with the embodiments.

[0044] Embodiment:

[0045] An improved ecological antibacterial yarn quality monitoring method based on machine vision of the present embodiment, such as Figure 1As shown, comprising: deploying a conduction assembly at the output end of the yarn production equipment, pushing the yarn output by the output end of the yarn production equipment based on the conduction assembly, collecting yarn image data during the transmission of the yarn in the conduction assembly;

[0046] The conduction assembly is integrated by a sleeve and a roller, the roller is symmetrically arranged inside the sleeve, the yarn output by the output end of the yarn production equipment enters the sleeve and contacts the surface of the roller, and the yarn is transmitted through the sleeve based on the traction of the output end of the yarn production equipment; The color of the inner wall of the sleeve and the surface of the roller is a solid color of an RGB color mode different from the color of the yarn, and the collection device for collecting yarn image data is arranged inside the sleeve. When the collection device collects yarn image data inside the sleeve each time, the collection of yarn image data is performed by applying an exposure mode;

[0047] Among them, the yarn image data is continuously collected based on a specified frequency, after each collection of yarn image data, the quality of the collected yarn image data is identified synchronously, and a yarn image data collection time node is inserted in the yarn image data collection frequency based on the quality of the yarn image data;

[0048] The quality identification logic of the yarn image data and the decision logic of the inserted yarn image data collection time node are represented as:

[0049] In the formula: Q is the quality performance value of the yarn image data; C and E are the definition of the clarity index and the exposure accuracy index; ω1 and ω2 are the weights; Q norr is a quality qualification threshold value defined based on the quality of the image data;

[0050] Among them, the weights ω1 and ω2 are positive numbers, and the sum is 1, the weight value is defined by the user end, and the weights ω1 and ω2 are initially set to 0.6 and 0.4. When formula (2) is established, the yarn image data is collected at a specified frequency, and when formula (2) is not established, the latest collected yarn image data collection timestamp is obtained, and the collection timestamp of the next collection of yarn image data is determined based on the collection frequency, which is recorded as a1 and a2. The inserted yarn image data collection time node is The calculation logic of C and E is represented as:

[0051]

[0052] In the formula: n is the total amount of pixels in the yarn image data; G x , G y is the gradient of the i-th pixel in the x and y directions calculated by the Sobel operator; is the average gray value after the yarn image is converted into a gray image; G ideal is the standard gray value;

[0053] Wherein, the standard gray value G ideal Customized by the system end user;

[0054] Through the logical formula calculation, the yarn image data quality is obtained, thereby providing further acquisition logic control support for the yarn image data acquisition logic.

[0055] The yarn image data acquisition time domain is set, and the acquired yarn image data is continuously acquired based on the acquisition time domain. When the acquisition time stamp corresponding to the earliest acquired yarn image data and the currently acquired yarn image data forms a time domain that meets the set yarn image data acquisition time domain, a yarn quality evaluation is performed once.

[0056] After the yarn image data acquisition operation based on the yarn image data acquisition time domain is completed, all acquired yarn image data is further traversed, and all yarn image data is sorted based on its acquisition time sequence. The yarn image data collected by inserting the yarn image data acquisition time node is identified based on the acquisition time sequence sorting.

[0057] In the yarn image similarity comparison stage, based on the identification result, the yarn image data collected according to the specified frequency is taken as image data group one, the yarn image data collected according to the inserted yarn image data acquisition time node is taken as image data group two, and the adjacent yarn image data collected according to the specified frequency and the yarn image data collected according to the inserted yarn image data acquisition time node are taken as image data group. The image data group is composed of several groups, and the similarity comparison is performed on the image data group and the image data group.

[0058] The similarity calculation formula of the image data group is:

[0059]

[0060] In the formula, M and N are the width and height of the yarn image data; F A (u,v), F B (u,v) is the representation result of image A and B in the frequency domain after two-dimensional discrete Fourier transform is performed on them respectively; is the complex conjugate of F B (u,v); (u,v) is the frequency domain coordinate;

[0061] In the formula, all yarn image data in image data group one and image data group two are calculated based on the above formula, and the results are averaged to obtain Acquiring yarn image data, segmenting yarn images in each yarn image data, comparing the similarity of each yarn image; when segmenting yarn images in the yarn image data, identifying the gray value interval of each pixel in the yarn image data through the image background gray value interval and the yarn image gray value interval defined by the user side, and based on the identification result, acquiring all the pixels in the yarn image gray value interval, and regarding the combination of the acquired pixels as the yarn image segmented in the yarn image data;

[0062] The similarity calculation formula of the image data set is:

[0063]

[0064] In the formula, c(x, y) is used to determine whether the edge directions of two images in the image data set at the corresponding pixel position (x, y) are consistent; e1(x, y), e2(x, y) are the edge intensity values of pixels (x, y) in two edge images obtained by using the Canny edge detection algorithm to perform edge detection on two images in the image data set; f(·) is a decision function:

[0065] Wherein, when the pixel (x, y) is an edge pixel, its edge intensity value is 1, otherwise its edge intensity value is 0, e1(x, y)+e2(x, y)>0, then f(·)=1, otherwise f(·)=0, and the similarity calculation results in a plurality of groups of image data are denoted as SIMM'1, SIMM'2, SIMM'3,...;

[0066] The value of c(x, y) is subject to:

[0067]

[0068] In the formula, θ1(x, y), θ2(x, y) are the quantized gradient directions of pixels (x, y) in two images in the image data set after edge detection; b is the number of direction intervals when quantizing the gradient direction to the direction interval;

[0069] Wherein, b is initially set to 8;

[0070] The above logical formula provides a specified calculation logic for similarity calculation between yarn image data, ensures stable completion of similarity calculation operation, and outputs the calculation result.

[0071] Acquiring the similarity comparison result of each yarn image, evaluating the production quality based on the comparison result, setting a yarn qualification determination threshold, comparing the set yarn qualification determination threshold with the yarn production quality evaluation result to determine whether the current production yarn is qualified; the yarn production quality evaluation logic is represented as:

[0072]

[0073] In the formula, P is a yarn production quality performance value; a, ε j are weights;

[0074] wherein the weights a, εj are greater than zero, and

[0075] In the present embodiment, through the execution of the method in the above embodiment, a higher real-time and higher precision quality monitoring system is brought to the ecological antibacterial yarn production line, the continuous image collection of each position on the output yarn is synchronized, the quality of the output yarn is comprehensively monitored, and the quality of the output yarn is ensured to be more stable.

[0076] Referring to Figure 2 as shown, referring to the label in the figure, the posture of the yarn in the transmission assembly is further shown, and the yarn image data collection is executed.

[0077] In summary, in the execution process of the method in the above embodiment, the yarn image data collection is provided with a stable collection environment by setting a specific transmission assembly, and the yarn produced by the yarn production equipment is brought to a global similarity analysis through a specified and continuous yarn image collection logic, so that the production yarn quality balance is evaluated based on the similarity analysis result. Compared with the existing precedent of applying image recognition technology to yarn quality evaluation, the precision and real-time performance are higher.

[0078] The above embodiments are only used to illustrate the technical solutions of the present application, but not to limit them; although the present application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that they can still modify the technical solutions recorded in the foregoing embodiments, or make equivalent replacement to part of the technical features; and these modifications or replacements will not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. An improved machine vision based eco-antibacterial yarn quality monitoring method, characterized in that, The application comprises the following steps: Deploying a conducting component at the output end of the yarn production equipment, pushing the yarn output by the output end of the yarn production equipment based on the conducting component, and collecting yarn image data during the transmission of the yarn in the conducting component; Setting a yarn image data acquisition time domain, continuously acquiring the collected yarn image data based on the acquisition time domain, and performing yarn quality evaluation once when the acquisition time stamp corresponding to the earliest acquired yarn image data and the currently acquired yarn image data forms a time domain that meets the set yarn image data acquisition time domain; Acquiring yarn image data, segmenting yarn images in each yarn image data, and comparing the similarity of each yarn image; Acquiring the similarity comparison result of each yarn image, evaluating the production quality based on the comparison result, setting a yarn qualification judgment threshold, comparing the set yarn qualification judgment threshold with the yarn production quality evaluation result, and determining whether the currently produced yarn is qualified; In the yarn image similarity comparison stage, based on the recognition result, the yarn image data collected according to the specified frequency is taken as image data group one, the yarn image data collected according to the inserted yarn image data acquisition time node is taken as image data group two, and the adjacent yarn image data collected according to the specified frequency and the yarn image data collected according to the inserted yarn image data acquisition time node are taken as image data groups. The image data groups are several groups, and similarity comparison is performed on the image data groups and the image data groups; The similarity calculation formula of the image data group is: ； wherein: is the width, height of the yarn image data; is the result of the two-dimensional discrete Fourier transform of the images A and B, respectively, in the frequency domain; is the complex conjugate of is the frequency domain coordinate;​ wherein all the yarn image data in the image data group one and the image data group two are calculated based on the above formula respectively, and the results are averaged, respectively denoted as ; The similarity calculation formula of the image data group is: ； In the formula: for determining whether the edge directions of two images in the image data set at the corresponding pixel position (x, y) are consistent; for edge detection of two images in the image data set using the Canny edge detection algorithm, the edge intensity value of pixel (x, y) in the two obtained edge images; for determining the function: Wherein, when the pixel (x, y) is an edge pixel, the edge intensity value thereof is 1, otherwise, the edge intensity value thereof is 0, > 0, then = 1, otherwise, then = 0, and the similarity calculation result in the image data of the groups is recorded as .

2. A machine vision based improved eco-antibacterial yarn quality monitoring method as claimed in claim 1 wherein, The conducting component is integrated by a sleeve and a roller, the roller is symmetrically arranged inside the sleeve, the yarn output by the output end of the yarn production equipment enters the sleeve, contacts the surface of the roller, and passes through the sleeve based on the traction force of the yarn transmission of the output end of the yarn production equipment; The inner wall of the sleeve and the surface of the roller are pure colors in RGB color mode that are different from the color of the yarn, the collection device for collecting yarn image data is arranged inside the sleeve, and the collection device applies an exposure mode to collect yarn image data when collecting yarn image data inside the sleeve each time; Wherein, the yarn image data is continuously collected based on the specified frequency, after collecting the yarn image data each time, the quality of the collected yarn image data is recognized synchronously, and a yarn image data acquisition time node is inserted in the yarn image data collection frequency based on the quality of the yarn image data.

3. A machine vision based improved eco-antibacterial yarn quality monitoring method as claimed in claim 2, wherein, The quality recognition logic of the yarn image data and the decision logic of the inserted yarn image data acquisition time node are represented as: ； In the formula: is a yarn image data quality performance value; is a sharpness index, an exposure accuracy index; is a weight; is a quality pass judgment threshold value defined based on an image data quality level. Wherein, the weight is a positive number, and the sum is 1, the weight value is defined by the user end, the weight is initially set to 0.6, 0.4, when formula (2) is established, the yarn image data is collected according to the specified frequency, when formula (2) is not established, the latest collected yarn image data collection time stamp is obtained, and the next yarn image data collection time stamp is determined based on the collection frequency, recorded as a1, a2, then the inserted yarn image data collection time node is .

4. A machine vision based improved eco-antibacterial yarn quality monitoring method as claimed in claim 3 wherein, The The computing logic is represented as: ； wherein: is the total amount of pixels in the yarn image data; is the gradient of the i-th pixel in the x and y direction calculated by the Sobel operator; is the average grey value after the yarn image is converted to a grey scale image; is the standard grey value; wherein the standard gray value defined by the system end user.

5. A machine vision based improved eco-antibacterial yarn quality monitoring method as claimed in claim 1, wherein, After the yarn image data is acquired based on the yarn image data acquisition time domain, all the acquired yarn image data is further traversed, all the yarn image data is sorted based on its collection time sequence, and the yarn image data collected through the inserted yarn image data acquisition time node in the yarn image data sorted based on the collection time sequence is recognized synchronously.

6. A machine vision based improved eco-antibacterial yarn quality monitoring method as claimed in claim 1 wherein, The values subject to: ； In the formula: is the quantized gradient direction of a pixel (x, y) in the image data set after edge detection; is the number of direction intervals when quantizing the gradient direction to the direction intervals. wherein is initially set to 8.

7. A machine vision based improved eco-antibacterial yarn quality monitoring method as claimed in claim 1 wherein, In the yarn image data, the yarn image is segmented, the image background gray value interval and the yarn image gray value interval are defined by the user, the gray value interval where each pixel gray value in the yarn image data is located is identified, based on the identification result, all the pixels in the yarn image gray value interval are obtained, and the combination of the obtained pixels is recorded as the yarn image segmented in the yarn image data.

8. A machine vision based improved eco-antibacterial yarn quality monitoring method as claimed in claim 1, wherein, The yarn production quality evaluation logic is represented as: ； In the formula: is a yarn production quality performance value; , is a weight; wherein the weights , are all greater than zero, , and .

Images