Cotton yarn color difference detection system based on image data

Abstract

The invention discloses a cotton yarn color difference detection system based on image data, belongs to the technical field of image processing, and is used for solving the problems of limitation of yarn color difference detection, such as incapability of accurately monitoring a yarn dyeing process image in real time in a large range and difficulty in effectively regulating and controlling a production process according to an image analysis result. Comprising an acquisition module, an image detection module, a matching module, a chromatic aberration analysis module, a repository, an alarm module and a chromatic aberration diagnosis module. The image detection module performs definition evaluation analysis by using an image processing technology, real-time monitoring and regulation of yarn dyeing quality are realized, the matching module is convenient for comprehensive detection of yarn chromatic aberration by splicing detection images, the chromatic aberration analysis module analyzes chromatic aberration deviation values of analysis images, and the yarn dyeing quality is improved. Chromatic aberration alarming and yarn quality grading are achieved, the chromatic aberration diagnosis module facilitates analysis of a parameter set with chromatic aberration, a worker can conduct targeted treatment conveniently, and the production efficiency is improved.

Description

Technical Field

[0001] The invention belongs to the technical field of image processing, and in particular relates to a cotton yarn color difference detection system based on image data. Background Art

[0002] In the textile industry, the quality inspection of cotton yarn is crucial to ensure the quality of the final product. As one of the key factors affecting the appearance quality of yarn, the accuracy of color difference inspection is directly related to whether the product meets the needs of the market and consumers. With the rapid development of the textile industry, the traditional manual inspection method can no longer meet the needs of modern production due to its low efficiency, strong subjectivity, and high labor intensity. In the process of winding after yarn dyeing, the yarn needs to be inspected for color difference to ensure the consistency of yarn color and the quality of the final product. At present, the color difference detection technologies on the market mainly include manual visual inspection and automatic inspection based on colorimeter. Manual visual inspection relies on the experience of the inspector and is easily affected by fatigue and subjective judgment, resulting in unstable inspection results. Although the automatic inspection method based on colorimeter can provide more objective measurement results, its detection range is limited and it is difficult to adapt to large-scale and continuous inspection needs.

[0003] Existing yarn color difference detection cannot accurately monitor the yarn dyeing process image in real time and over a large range, and regulate the yarn dyeing production according to the image analysis results. At the same time, it is not convenient to grade the quality of the existing yarn after production, and there is a lack of guarantee for product consistency. For this reason, we propose a cotton yarn color difference detection system based on image data. Summary of the invention

[0004] The purpose of the present invention is to provide a cotton yarn color difference detection system based on image data to solve the problems raised in the above background technology. To achieve the above purpose, the present invention provides the following technical solutions: a cotton yarn color difference detection system based on image data, comprising an acquisition module, an image detection module, a matching module, a color difference analysis module, a storage library, an alarm module, and a color difference diagnosis module; Acquisition module: The image information of the wound yarn is collected in real time through multiple groups of high-speed cameras, and the yarn detection images taken by the high-speed cameras in real time are obtained and sent to the matching module. At the same time, according to the preset acquisition interval, a group of detection images are regularly extracted from each group of high-speed cameras and sent to the image analysis module; Image detection module: obtains the image clarity evaluation value of the detection image, and adjusts the environmental parameters according to the image clarity evaluation value of the detection image; Matching module: matches and stitches multiple groups of detections sent by the acquisition module at the same time into a complete analysis image; Color difference analysis module: perform color difference analysis on the analysis image, calculate the color difference deviation value of the analysis image, and judge the degree of yarn color difference according to the color difference deviation value of the analysis image, and make alarms and quality ratings; Repository: stores test images, standard images, environmental parameter sets, color difference deviation value diagnostic parameter sets and other data; Alarm module: When it detects that the image clarity is insufficient or the color difference deviation value exceeds the preset threshold, it will make a voice announcement and control the relevant equipment to stop working; Color difference diagnosis module: establish color difference deviation value diagnosis model and color deviation total product diagnosis model to diagnose the cause parameters of color difference.

[0005] Preferably, the acquisition module works in the following manner: A plurality of groups of high-speed cameras are arranged between the exit of the yarn dyeing machine and the winding machine and are synchronously controlled to synchronously monitor all areas that need to be monitored for yarn winding, and the yarn image taken in real time by the high-speed camera is obtained and marked as a detection image i, where i represents the number of the detection image, i=1, 2, 3, ..., n; the detection images collected by the plurality of groups of high-speed cameras are sent to the matching module, and the image clarity detection interval is preset at the same time. According to the preset collection interval, a group of detection images is regularly extracted from each group of high-speed cameras and sent to the image analysis module.

[0006] Preferably, the working mode of the image detection module is: Mark some parameter objects with fixed colors in the detection image as detection objects, extract the detection object image in the detection image, convert the detection object image from the RGB color space to the HSV color space, evenly set a number of identification points in the detection object image, and for each identification point, obtain its H, S and V channel values ​​in the HSV color space. For all identification points in the detection object image, calculate the sum of the H, S, and V channel values ​​of each channel respectively, and then calculate the average value of each channel value as the H, S, and V channel value of the detection object image; Match the image of the object to be detected with all the standard images of the object to be detected in the storage library to obtain the standard image of the corresponding object to be detected, extract the H, S, and V channel values ​​of the standard image of the corresponding object to be detected, calculate the deviation values ​​between the H, S, and V channel values ​​of the image of the object to be detected and the H, S, and V channel values ​​of the corresponding standard image, and mark them as the detection image H channel deviation value HPi, the detection image S channel deviation value SPi, and the detection image V channel deviation value VPi, and use the formula: TXi=HPi*a1+SPi*a2+VPi*a3 to obtain the image clarity evaluation value TXi, where a1, a2, and a3 are preset weight coefficients; The image clarity evaluation value TXi is compared with the preset image clarity deviation threshold. If the image clarity evaluation value TXi is larger, the image clarity evaluation value TXi is substituted into the environmental parameter model to obtain the environmental parameter set that affects the image clarity. The environmental parameter set that affects the image clarity and the current time are sent to the repository. At the same time, an environmental adjustment instruction is generated and sent to the alarm module. After receiving the environmental adjustment instruction, the alarm module will broadcast the environmental adjustment voice. After hearing the environmental adjustment voice broadcast, the staff will obtain the environmental parameter set that affects the image clarity from the repository according to the start time of the environmental adjustment voice broadcast, and perform corresponding processing according to the environmental parameter set that affects the image clarity.

[0007] Preferably, the process of establishing the environmental parameter model is: A series of environmental parameter data that may affect image clarity were collected from historical records, including light intensity, air humidity, ambient temperature, lens dust content, and technical parameters of the camera, such as aperture size, focal length, and ISO setting. The Pearson correlation coefficient statistical method was used to quantitatively analyze the correlation between each environmental parameter and the image clarity evaluation value TXi. Based on the results of the correlation analysis, the top three environmental parameters with the highest correlation were selected as the input feature set of the model to construct a neural network model. The collected data was used to train the neural network model. The performance of the model was optimized by adjusting the network weights and biases. After training and adjustment, the environmental parameter model was obtained.

[0008] Preferably, the matching module works as follows: The edge feature points of the detection images sent by the acquisition module at the same time are extracted, the feature points in the detection images taken by different high-speed cameras are matched, and the spatial correspondence between the detection images is established. According to the spatial correspondence between the detection images, the geometric transformation relationship between the detection images is calculated. The detection images sent by the acquisition module at the same time are aligned into a common reference coordinate system by using the geometric transformation relationship between the detection images. The overlapping areas of the detection images are fused at the pixel level. Finally, the aligned and fused multiple groups of detection images are spliced ​​into a complete large-range image, marked as analysis image t, and the analysis image t is sent to the color difference analysis module; wherein t represents the time number of the analysis image, t=1, 2, 3, ..., n.

[0009] Preferably, the color difference analysis module works as follows: Evenly set a number of color difference detection points in the analysis image, perform RGB color recognition on each color difference detection point, obtain the RGB color value of each color difference detection point, compare the RGB color value of each color difference detection point with the RGB color value of the target yarn, calculate the difference, mark the color difference detection point with a positive difference as a brighter detection point, and mark the color difference detection point with a negative difference as a darker detection point, set the brighter detection point and the darker detection point to have a preset RGB value deviation interval, match each brighter detection point with the preset deviation RGB value interval of the brighter detection point, and match each darker detection point with the preset RGB value deviation interval of the darker detection point; The bright detection points that are not in the preset RGB value deviation interval of the bright detection points are marked as bright secondary points, and the number of bright secondary points PLc t is counted, where c represents the preference of the color difference detection point, c=1, 2, 3, ..., n; at the same time, the distance values ​​of all bright secondary points from the preset RGB value deviation interval of the bright detection points are obtained and marked as the total bright value PLZc t. Similarly, the number of dark secondary points PAc t and the total dark value PAZc t are obtained, using the formula: , get the color difference deviation value FPZ of the analyzed image t , where w1 and w2 are preset weight coefficients, and the color difference deviation value FPZ of the analyzed image is t Compare with the preset color difference deviation alarm value of the analysis image. If the color difference deviation value FPZ of the analysis image is t If the color difference is larger, a yarn color difference warning instruction is generated and sent to the alarm module. When the alarm module receives the yarn color difference deviation warning instruction, it broadcasts the color difference voice and controls the yarn winding machine and the yarn dyeing machine to stop working. At the same time, the color difference deviation value FPZ of the image is analyzed. t When it is large, the color difference deviation value FPZ of the analyzed image will be t Sent to the color difference diagnosis module.

[0010] Preferably, if the color difference deviation value of the analyzed image is small, the processing method is: If the color difference deviation value FPZ of the image is analyzed t If the color difference deviation value of the analyzed image is smaller, the color difference deviation value of the analyzed image is substituted into the curve diagram between the color difference deviation value of the analyzed image and time, the values ​​of the color difference deviation value of the analyzed image at each time point are connected to obtain the image color deviation threshold line, a preset analysis image color deviation threshold is drawn on the image color deviation threshold line, the area where the image color deviation threshold line is located above the preset analysis image color deviation threshold and enclosed by it is marked as the color deviation increase area, all the color deviation increase areas are counted and marked as the color deviation increase total area, the color deviation increase total area alarm threshold is preset, and the color deviation increase total area is compared with the preset color deviation increase total area alarm threshold; When the total product of color deviation is greater than the preset alarm threshold of total product of color deviation, a yarn color difference deviation warning instruction will be generated and sent to the alarm module. When the alarm module receives the yarn color difference deviation warning, it will broadcast the yarn color difference increase voice, and control the yarn winding machine and the yarn dyeing machine to stop working, and send the total product of color deviation to the color difference diagnosis module. When the yarn winding machine ends and the total product of color deviation is still less than the preset total product of color deviation alarm threshold, three yarn quality levels are set, and the total product of color deviation is matched with the total product of color deviation intervals corresponding to the three set yarn quality levels to obtain the quality level of the finished yarn.

[0011] Preferably, the color difference diagnosis module works as follows: Taking the establishment of the environmental parameter model as the reference model establishment method, the color difference deviation value diagnosis model and the color deviation total product diagnosis model of the analysis image are established similarly; the color difference deviation value or the color deviation total product of the analysis image received by the color difference diagnosis module is substituted into the corresponding diagnosis model to obtain the color difference deviation value diagnosis parameter set or the color deviation total product diagnosis parameter set of the analysis image, and the color difference deviation value diagnosis parameter set or the color deviation total product diagnosis parameter set is sent to the storage library. The staff can retrieve the color difference deviation value diagnosis parameter set or the color deviation total product diagnosis parameter set from the storage library according to the start time of the color difference voice broadcast or the start time of the yarn color difference increase voice broadcast to process the yarn color difference.

[0012] Compared with the prior art, the present invention has the following beneficial effects: (1) The cotton yarn color difference detection system based on image data collects the image information of the wound yarn in real time through multiple groups of high-speed cameras to obtain the detection image in the yarn production process, analyzes the image clarity evaluation value of the detection image, and obtains the clarity of the detection image. If it is not clear, the image clarity evaluation value is substituted into the environmental parameter model to obtain the environmental parameter set that affects the image clarity, and then optimizes the environmental parameters according to the environmental parameter set that affects the image clarity to ensure the clarity of image acquisition and provide a more accurate analysis image for subsequent yarn detection and analysis. At the same time, the images taken by multiple high-definition cameras can be spliced ​​through the matching module, which is helpful for comprehensive detection of yarn color difference.

[0013] (2) The color difference detection system for cotton yarn based on image data, the color difference analysis module analyzes the color difference deviation value of the analysis image and compares it with the corresponding preset alarm value, realizing two alarm mechanisms: color difference instantaneous alarm and accumulation alarm. The color difference instantaneous alarm targets the color difference deviation at a single time point. When the color difference deviation value of the analysis image exceeds the threshold, the alarm is immediately triggered and production is suspended to reduce unqualified products; the accumulation alarm focuses on the cumulative effect of color difference over time. If the total product of color deviation exceeds the set threshold, the alarm will also be triggered to ensure long-term quality control. Finally, the yarn quality is graded according to the accumulation of the total product of color deviation, so as to effectively distinguish and ensure the quality of yarn products.

[0014] (3) The cotton yarn color difference detection system based on image data can facilitate the diagnosis of the parameter set that causes the color difference through the color difference diagnosis module. According to the parameter set that causes the color difference, the staff can perform targeted color difference processing, thereby improving the efficiency of yarn production adjustment and thus accelerating the overall efficiency of yarn production. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] Figure 1 It is a flowchart of the present invention; Figure 2 A curve diagram showing the color deviation threshold values ​​of the present invention; DETAILED DESCRIPTION

[0016] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0017] See also Figure 1 - Figure 2 Embodiment 1:

[0018] The present invention provides a cotton yarn color difference detection system based on image data, including an acquisition module, an image detection module, a matching module, a color difference analysis module, a storage library, an alarm module, and a color difference diagnosis module; The acquisition module uses multiple sets of high-speed cameras to collect the image information of the wound yarn in real time. The specific process is as follows: A plurality of groups of high-speed cameras are arranged between the exit of the yarn dyeing machine and the winding machine and are synchronously controlled to synchronously monitor all areas that need to be monitored for yarn winding, obtain the yarn image taken by the high-speed camera in real time, and mark it as a detection image i, where i represents the number of the detection image, i=1, 2, 3, ..., n; by sending the detection images collected by the plurality of groups of high-speed cameras to the matching module, and presetting the image clarity detection interval, a group of detection images is regularly extracted from each group of high-speed cameras according to the preset collection interval, and sent to the image analysis module; The image detection module performs image clarity detection on each set of detection images extracted by the high-speed camera within the acquisition interval, specifically: Mark some parameter objects with fixed colors in the detection image as detection objects, extract the detection object image in the detection image, convert the detection object image from the RGB color space to the HSV color space, evenly set a number of identification points in the detection object image, and for each identification point, obtain its H (hue), S (saturation) and V (brightness) channel values ​​in the HSV color space. For all identification points in the detection object image, calculate the sum of each channel (H, S, V) value, and then calculate the average value of each channel value as the H, S, V channel value of the detection object image; The storage library stores standard images of all detection objects. The detection object image is matched with all the detection object standard images in the storage library to obtain the standard image of the corresponding detection object, and the H, S, and V channel values ​​of the standard image of the corresponding detection object are extracted. The deviation values ​​between the H, S, and V channel values ​​of the detection object image and the H, S, and V channel values ​​of the corresponding standard image are calculated and marked as the detection image H channel deviation value HPi, the detection image S channel deviation value SPi, and the detection image V channel deviation value VPi. The formula: TXi=HPi*a1+SPi*a2+VPi*a3 is used to obtain the image clarity evaluation value TXi, wherein a1, a2, and a3 are preset weight coefficients; wherein the larger the image clarity evaluation value TXi is, the lower the image clarity is; The image clarity evaluation value TXi is compared with the preset image clarity deviation threshold. If the image clarity evaluation value TXi is larger, it indicates that the image clarity is insufficient. Then the image clarity evaluation value TXi is substituted into the environmental parameter model to obtain the environmental parameter set affecting the image clarity. The environmental parameter set affecting the image clarity and the current time are sent to the storage library. At the same time, an environmental adjustment instruction is generated and sent to the alarm module. After receiving the environmental adjustment instruction, the alarm module will make an environmental adjustment voice broadcast. After hearing the environmental adjustment voice broadcast, the staff can obtain the environmental parameter set affecting the image clarity from the storage library according to the start time of the environmental adjustment voice broadcast, and perform corresponding processing according to the environmental parameter set affecting the image clarity to improve the clarity of the high-speed camera image acquisition. If the image clarity evaluation value TXi is smaller, it is considered that the clarity of the current high-speed camera image acquisition is satisfactory, and no additional measures are required. The process of establishing the environmental parameter model is as follows: a series of environmental parameter data that may affect image clarity are collected from historical records, including light intensity, air humidity, ambient temperature, lens dust content, and camera technical parameters such as aperture size, focal length, and ISO setting, and the collected historical data of environmental parameters are preprocessed; Based on the preprocessed data, the Pearson correlation coefficient statistical method is used to quantitatively analyze the correlation between each environmental parameter and the image clarity evaluation value TXi. Based on the results of the correlation analysis, the top three environmental parameters with the highest correlation are selected as the input feature set of the model to construct a neural network model. The neural network model is then trained using the collected data. The performance of the model is optimized by adjusting the network weights and biases. After training and adjustment, the environmental parameter model is obtained.

[0019] It should be noted that, through the image detection module, the clarity of the detection images collected by multiple groups of high-speed cameras in the acquisition module can be detected to ensure the clarity of the detection images collected by the acquisition module, provide a clearer detection image for the subsequent color difference analysis module, and thus improve the accuracy of the analysis of the color difference analysis module. Embodiment 2:

[0020] Matching module: Match and stitch the detection images sent by the acquisition module at the same time into a complete large-scale image, which covers the monitoring area of ​​all high-speed cameras. The specific process is as follows: The detection images sent by the acquisition module at the same time are preprocessed, including denoising, distortion correction, contrast enhancement, etc., to improve the image quality. The edge feature points of the detection images are extracted, wherein the edge feature points include edge texture and edge color. After the edge feature points of the detection images are extracted, the SIFT feature matching algorithm is used to match the feature points in the detection images taken by different high-speed cameras, and the spatial correspondence between the detection images is established. According to the spatial correspondence between the detection images, the geometric transformation relationship between the detection images is calculated, and the geometric transformation relationship includes translation, rotation and scaling. The detection images sent by the acquisition module at the same time are aligned into a common reference coordinate system by using the geometric transformation relationship between the detection images, and the overlapping areas of the detection images are fused at the pixel level to eliminate the splicing traces. Finally, the aligned and fused multiple groups of detection images are spliced ​​into a complete large-range image, and marked as analysis image t, and the analysis image t is sent to the color difference analysis module; wherein t represents the time number of the analysis image, t=1, 2, 3, ..., n; wherein the analysis image covers the monitoring area of ​​all high-definition cameras.

[0021] It should be noted that the detection images taken by multiple groups of high-speed cameras are spliced ​​together through the matching module to form a more comprehensive yarn production analysis image, which is convenient for the subsequent color difference analysis module to conduct comprehensive detection of the yarn on the entire production line.

[0022] The color difference analysis module performs color difference analysis on the analysis image to obtain the color difference deviation value of the analysis image. Then, according to the size of the color difference deviation value of the analysis image, it is judged whether the degree of yarn color difference needs to be alarmed. At the same time, according to the corresponding relationship between the color difference deviation value of the analysis image and time, the yarn quality is rated. The specific process is as follows: Evenly set a number of color difference detection points in the analysis image, perform RGB color recognition on each color difference detection point, obtain the RGB color value of each color difference detection point, compare the RGB color value of each color difference detection point with the RGB color value of the target yarn, calculate the difference, mark the color difference detection point with a positive difference as a brighter detection point, and mark the color difference detection point with a negative difference as a darker detection point, set the brighter detection point and the darker detection point to have a preset RGB value deviation interval, match each brighter detection point with the preset deviation RGB value interval of the brighter detection point, and match each darker detection point with the preset RGB value deviation interval of the darker detection point; The bright detection points that are not in the preset RGB value deviation interval of the bright detection points are marked as bright secondary points, and the number of bright secondary points PLc t is counted, where c represents the preference of the color difference detection point, c=1, 2, 3, ..., n; at the same time, the distance values ​​of all bright secondary points from the preset RGB value deviation interval of the bright detection points are obtained and marked as the total bright value PLZc t. Similarly, the number of dark secondary points PAc t and the total dark value PAZc t are obtained, using the formula: , get the color difference deviation value FPZ of the analyzed image t , where w1 and w2 are preset weight coefficients, and the color difference deviation value FPZ of the analyzed image is t Compare with the preset color difference deviation alarm value of the analysis image. If the color difference deviation value FPZ of the analysis image is t If the color difference is larger, a yarn color difference warning instruction is generated and sent to the alarm module. When the alarm module receives the yarn color difference deviation warning instruction, it broadcasts the color difference voice and controls the yarn winding machine and the yarn dyeing machine to stop working. At the same time, the color difference deviation value FPZ of the image is analyzed. t When it is large, the color difference deviation value FPZ of the analyzed image will be t Send to the color difference diagnosis module; If the color difference deviation value FPZ of the image is analyzed t If the color difference deviation value of the analysis image is smaller, the color difference deviation value of the analysis image is substituted into the curve diagram between the color difference deviation value of the analysis image and time, such as Figure 2 , connect the values ​​of the color difference deviation values ​​of the analyzed image at each time point to obtain an image color deviation threshold line, draw a preset analyzed image color deviation threshold on the image color deviation threshold line, mark the area enclosed by the image color deviation threshold line above the preset analyzed image color deviation threshold as a color deviation increase area, count all color deviation increase areas, and mark them as a color deviation total product, wherein the color deviation total product may increase with the increase of time, preset a color deviation total product alarm threshold, and compare the color deviation total product with the preset color deviation total product alarm threshold; When the total product of color deviation is greater than the preset alarm threshold of total product of color deviation, a yarn color difference deviation warning instruction will be generated and sent to the alarm module. When the alarm module receives the yarn color difference deviation warning, it will broadcast the yarn color difference increase voice, and control the yarn winding machine and the yarn dyeing machine to stop working, and send the total product of color deviation to the color difference diagnosis module. When the yarn winding machine ends, if the total product of color deviation is still less than the preset alarm threshold of total product of color deviation, three yarn quality levels are set, namely, first-level yarn quality level, second-level yarn quality level, and third-level yarn quality level, wherein each yarn quality level corresponds to an interval of total product of color deviation, wherein the higher the level, the larger the interval of total product of color deviation corresponding to the yarn quality level, the greater the yarn color difference, and the worse the quality, and the total product of color deviation is matched with the interval of total product of color deviation corresponding to the three set yarn quality levels to obtain the quality level of the finished yarn.

[0023] It should be noted that the color difference analysis module analyzes the color difference deviation value of the analysis image and compares it with the corresponding preset alarm value, thereby realizing two alarm mechanisms: color difference instantaneous alarm and accumulation alarm. The color difference instantaneous alarm targets the color difference deviation at a single time point. When the color difference deviation value of the analysis image exceeds the threshold, the alarm is immediately triggered and production is suspended to reduce unqualified products; the accumulation alarm focuses on the cumulative effect of color difference over time. If the total product of color deviation exceeds the set threshold, the alarm will also be triggered to ensure long-term quality control. Finally, the yarn is graded according to the accumulation of the total product of color deviation, thereby effectively distinguishing and ensuring the quality of yarn products.

[0024] The color difference diagnosis module works as follows: Taking the establishment of the environmental parameter model as the reference model establishment method, the color difference deviation value diagnosis model and the color deviation total product diagnosis model of the analysis image are established in the same way; The color difference deviation value or the total color deviation product of the analysis image received by the color difference diagnosis module is substituted into the corresponding diagnosis model to obtain the color difference deviation value diagnosis parameter set or the total color deviation product diagnosis parameter set of the analysis image, and the color difference deviation value diagnosis parameter set or the total color deviation product diagnosis parameter set is sent to the storage library. The staff can retrieve the color difference deviation value diagnosis parameter set or the total color deviation product diagnosis parameter set from the storage library according to the start time of the color difference voice broadcast or the start time of the yarn color difference increase voice broadcast to process the yarn color difference.

[0025] It should be noted that the color difference diagnosis module can facilitate the diagnosis of the parameter set that causes the color difference. According to the parameter set that causes the color difference, the staff can perform targeted color difference processing, thereby improving the efficiency of yarn production adjustment and speeding up the overall efficiency of yarn production.

[0026] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A cotton yarn color difference detection system based on image data, comprising an acquisition module, an image detection module, a matching module, a color difference analysis module, a storage library, an alarm module, and a color difference diagnosis module, characterized in that: Acquisition module: The image information of the wound yarn is collected in real time through multiple groups of high-speed cameras, and the yarn detection images taken by the high-speed cameras in real time are obtained and sent to the matching module. At the same time, according to the preset acquisition interval, a group of detection images are regularly extracted from each group of high-speed cameras and sent to the image analysis module; Image detection module: obtains the image clarity evaluation value of the detection image, and adjusts the environmental parameters according to the image clarity evaluation value of the detection image; Matching module: matches and stitches multiple groups of detections sent by the acquisition module at the same time into a complete analysis image; Color difference analysis module: perform color difference analysis on the analysis image, calculate the color difference deviation value of the analysis image, and judge the degree of yarn color difference according to the color difference deviation value of the analysis image, and make alarms and quality ratings; Repository: stores test images, standard images, environmental parameter sets, color difference deviation value diagnostic parameter sets and other data; Alarm module: When it detects that the image clarity is insufficient or the color difference deviation value exceeds the preset threshold, it will make a voice announcement and control the relevant equipment to stop working; Color difference diagnosis module: establish color difference deviation value diagnosis model and color deviation total product diagnosis model to diagnose the cause parameters of color difference.

2. The cotton yarn color difference detection system based on image data according to claim 1 is characterized in that: The acquisition module works as follows: A plurality of high-speed cameras are arranged between the exit of the yarn dyeing machine and the winding machine and are synchronously controlled to synchronously monitor all the areas that need to be monitored for yarn winding, and the yarn image taken by the high-speed camera in real time is obtained and marked as detection image i, where i represents the number of the detection image, i=1, 2, 3, ..., n; The detection images collected by multiple groups of high-speed cameras are sent to the matching module, and the image clarity detection interval is preset. According to the preset collection interval, a group of detection images is regularly extracted from each group of high-speed cameras and sent to the image analysis module.

3. The cotton yarn color difference detection system based on image data according to claim 2 is characterized in that: The image detection module works as follows: Mark some parameter objects with fixed colors in the detection image as detection objects, extract the detection object image in the detection image, convert the detection object image from the RGB color space to the HSV color space, evenly set a number of identification points in the detection object image, and for each identification point, obtain its H, S and V channel values ​​in the HSV color space. For all identification points in the detection object image, calculate the sum of the H, S, and V channel values ​​of each channel respectively, and then calculate the average value of each channel value as the H, S, and V channel value of the detection object image; Match the image of the object to be detected with all the standard images of the object to be detected in the storage library to obtain the standard image of the corresponding object to be detected, extract the H, S, and V channel values ​​of the standard image of the corresponding object to be detected, calculate the deviation values ​​between the H, S, and V channel values ​​of the image of the object to be detected and the H, S, and V channel values ​​of the corresponding standard image, and mark them as the detection image H channel deviation value HPi, the detection image S channel deviation value SPi, and the detection image V channel deviation value VPi, and use the formula: TXi=HPi*a1+SPi*a2+VPi*a3 to obtain the image clarity evaluation value TXi, where a1, a2, and a3 are preset weight coefficients; The image clarity evaluation value TXi is compared with the preset image clarity deviation threshold. If the image clarity evaluation value TXi is larger, the image clarity evaluation value TXi is substituted into the environmental parameter model to obtain the environmental parameter set that affects the image clarity. The environmental parameter set that affects the image clarity and the current time are sent to the repository. At the same time, an environmental adjustment instruction is generated and sent to the alarm module. After receiving the environmental adjustment instruction, the alarm module will broadcast the environmental adjustment voice. After hearing the environmental adjustment voice broadcast, the staff will obtain the environmental parameter set that affects the image clarity from the repository according to the start time of the environmental adjustment voice broadcast, and perform corresponding processing according to the environmental parameter set that affects the image clarity.

4. The cotton yarn color difference detection system based on image data according to claim 3 is characterized in that: The process of establishing the environmental parameter model is as follows: A series of environmental parameter data that may affect image clarity were collected from historical records, including light intensity, air humidity, ambient temperature, lens dust content, and technical parameters of the camera, such as aperture size, focal length, and ISO setting. The Pearson correlation coefficient statistical method was used to quantitatively analyze the correlation between each environmental parameter and the image clarity evaluation value TXi. Based on the results of the correlation analysis, the top three environmental parameters with the highest correlation were selected as the input feature set of the model to construct a neural network model. The collected data was used to train the neural network model. The performance of the model was optimized by adjusting the network weights and biases. After training and adjustment, the environmental parameter model was obtained.

5. The cotton yarn color difference detection system based on image data according to claim 4 is characterized in that: The matching module works as follows: The edge feature points of the detection images sent by the acquisition module at the same time are extracted, the feature points in the detection images taken by different high-speed cameras are matched, and the spatial correspondence between the detection images is established. According to the spatial correspondence between the detection images, the geometric transformation relationship between the detection images is calculated. The detection images sent by the acquisition module at the same time are aligned into a common reference coordinate system by using the geometric transformation relationship between the detection images. The overlapping areas of the detection images are fused at the pixel level. Finally, the aligned and fused multiple groups of detection images are spliced ​​into a complete large-range image, marked as analysis image t, and the analysis image t is sent to the color difference analysis module; wherein t represents the time number of the analysis image, t=1, 2, 3, ..., n.

6. The cotton yarn color difference detection system based on image data according to claim 1, characterized in that: The color difference analysis module works as follows: Evenly set a number of color difference detection points in the analysis image, perform RGB color recognition on each color difference detection point, obtain the RGB color value of each color difference detection point, compare the RGB color value of each color difference detection point with the RGB color value of the target yarn, calculate the difference, mark the color difference detection point with a positive difference as a brighter detection point, and mark the color difference detection point with a negative difference as a darker detection point, set the brighter detection point and the darker detection point to have a preset RGB value deviation interval, match each brighter detection point with the preset deviation RGB value interval of the brighter detection point, and match each darker detection point with the preset RGB value deviation interval of the darker detection point; The bright detection points that are not in the preset RGB value deviation interval of the bright detection points are marked as bright secondary points, and the number of bright secondary points PLc t is counted, where c represents the preference of the color difference detection point, c=1, 2, 3, ..., n; at the same time, the distance values ​​of all bright secondary points from the preset RGB value deviation interval of the bright detection points are obtained and marked as the total bright value PLZc t. Similarly, the number of dark secondary points PAc t and the total dark value PAZc t are obtained, using the formula: , get the color difference deviation value FPZ of the analyzed image t , where w1 and w2 are preset weight coefficients, and the color difference deviation value FPZ of the analyzed image is t Compare with the preset color difference deviation alarm value of the analysis image. If the color difference deviation value FPZ of the analysis image is t If the color difference is larger, a yarn color difference warning instruction is generated and sent to the alarm module. When the alarm module receives the yarn color difference deviation warning instruction, it broadcasts the color difference voice and controls the yarn winding machine and the yarn dyeing machine to stop working. At the same time, the color difference deviation value FPZ of the image is analyzed. t When it is large, the color difference deviation value FPZ of the analyzed image will be t Sent to the color difference diagnosis module.

7. The cotton yarn color difference detection system based on image data according to claim 6 is characterized in that: If the color difference deviation value of the analyzed image is small, the processing method is: If the color difference deviation value FPZ of the image is analyzed t If the color difference deviation value of the analyzed image is smaller, the color difference deviation value of the analyzed image is substituted into the curve diagram between the color difference deviation value of the analyzed image and time, the values ​​of the color difference deviation value of the analyzed image at each time point are connected to obtain the image color deviation threshold line, a preset analysis image color deviation threshold is drawn on the image color deviation threshold line, the area where the image color deviation threshold line is located above the preset analysis image color deviation threshold and enclosed by it is marked as the color deviation increase area, all the color deviation increase areas are counted and marked as the color deviation increase total area, the color deviation increase total area alarm threshold is preset, and the color deviation increase total area is compared with the preset color deviation increase total area alarm threshold; When the total product of color deviation is greater than the preset alarm threshold of total product of color deviation, a yarn color difference deviation warning instruction will be generated and sent to the alarm module. When the alarm module receives the yarn color difference deviation warning, it will broadcast the yarn color difference increase voice, and control the yarn winding machine and the yarn dyeing machine to stop working, and send the total product of color deviation to the color difference diagnosis module. When the yarn winding machine ends and the total product of color deviation is still less than the preset total product of color deviation alarm threshold, three yarn quality levels are set, and the total product of color deviation is matched with the total product of color deviation intervals corresponding to the three set yarn quality levels to obtain the quality level of the finished yarn.

8. The cotton yarn color difference detection system based on image data according to claim 7, characterized in that: The color difference diagnosis module works as follows: Taking the establishment of the environmental parameter model as the reference model establishment method, the color difference deviation value diagnosis model and the color deviation total product diagnosis model of the analysis image are established similarly; the color difference deviation value or the color deviation total product of the analysis image received by the color difference diagnosis module is substituted into the corresponding diagnosis model to obtain the color difference deviation value diagnosis parameter set or the color deviation total product diagnosis parameter set of the analysis image, and the color difference deviation value diagnosis parameter set or the color deviation total product diagnosis parameter set is sent to the storage library. The staff can retrieve the color difference deviation value diagnosis parameter set or the color deviation total product diagnosis parameter set from the storage library according to the start time of the color difference voice broadcast or the start time of the yarn color difference increase voice broadcast to process the yarn color difference.

Images