CN118417184B - Intelligent textile automatic sorting system based on machine vision - Google Patents

Abstract

The invention discloses an intelligent textile automatic sorting system based on machine vision, which reduces manual input errors and quickens sorting speed by automatically identifying textile labels, collects image information of textiles and analyzes the image information when sorting the textiles to a corresponding target workshop, comprehensively evaluates stains, defects and sutures to generate flaw estimation values, provides comprehensive data support for quality control of the textiles, improves accuracy of quality evaluation, further generates bar codes comprising flaw grade information according to the flaw estimation values of the corresponding textiles, attaches the bar codes to the corresponding textiles, and continues to convey the bar codes after the attachment is finished until the bar codes are conveyed to the corresponding target workshop of the textiles, thereby solving the problem that the quality of the textiles cannot be evaluated based on quality requirements of the textiles in the prior art, and packages and sends evaluation results and the textiles to the corresponding sorting target workshop.

Description

An intelligent textile automatic sorting system based on machine vision

Technical Field

The present invention relates to the technical field of textile sorting, and in particular to an intelligent automatic textile sorting system based on machine vision.

Background Art

With the rapid development of industrial automation, the demand for automated sorting in the textile manufacturing industry is growing. Traditional manual sorting is not only inefficient but also prone to errors and cannot meet the needs of large-scale production. Therefore, automatic sorting is needed to replace manual work to meet production needs.

However, the existing automatic textile sorting system still has the following deficiencies in practical application:

During the sorting process of textiles, it is impossible to evaluate the production quality based on the quality requirements of textiles, and the evaluation results and the textiles cannot be packaged and sent to the corresponding sorting target workshop. The degree of intelligence is low and the subsequent quality assessment task through manpower is heavy, which easily leads to the accumulation of goods.

To this end, an intelligent textile automatic sorting system based on machine vision is introduced.

Summary of the invention

In view of this, the present invention provides an intelligent automatic textile sorting system based on machine vision to solve the problems raised by the above background technology.

The purpose of the present invention can be achieved by the following technical solution: an intelligent automatic textile sorting system based on machine vision, comprising:

Classification decision module: When the textile moves along the conveyor belt and passes through the RFID reading area, the RFID tag on each textile is scanned and identified. After identification, the information in the RFID tag of each textile is parsed to determine the target workshop for each textile. According to the determined target workshop, the execution device is controlled to transfer the corresponding textile from the main conveyor belt to the branch conveyor belt of the target workshop.

Image analysis module: when each piece of textile is conveyed through the corresponding branch conveyor belt, the image acquisition device set on each branch conveyor belt collects the image information of each piece of textile, pre-processes and analyzes each group of image information, extracts the color and pattern information matching the label from the analysis results of each group of image information, and matches and verifies the color verification information and pattern information of the corresponding textile with the color and pattern information in the label respectively. If the verification is passed, the image information of the corresponding textile is sent to the quality assessment module;

Quality assessment module: receiving image information of the corresponding textile, and analyzing it to obtain a defect estimation value Defal of the corresponding textile; generating a barcode including defect grade information based on the defect estimation value Defal of the corresponding textile, attaching the barcode to the corresponding textile, and continuing to transport it after the attachment is completed until it is transported to the target workshop of the corresponding textile;

Abnormality detection module: monitors the sorting process in real time, and identifies the specific type of abnormality through the set self-check mechanism when an abnormality occurs. After the identification is completed, the maintenance signal corresponding to the specific type of abnormality is triggered, and the corresponding steps are executed.

In some embodiments, image information of the corresponding textile is received and analyzed, specifically:

First, the image is enhanced, and after the enhancement, the stain area in the image is separated from the background, and classified according to the characteristics of the stain area, including oil stain characteristics, ink characteristics and dye penetration characteristics, and a weight coefficient is set for the oil stain characteristics, ink characteristics and dye penetration characteristics respectively; the number of pixels of each stain area in the image is counted, and the image is converted based on the resolution to obtain each group of stain areas of the corresponding textile; the stain areas of the oil stain characteristics are accumulated as the oil stain area of the corresponding textile; the stain areas of the ink characteristics are accumulated as the ink area of the corresponding textile; the stain areas of the dye penetration characteristics are accumulated as the penetration area of the corresponding textile; the weight coefficients of each feature are extracted, the oil stain area, ink area and penetration area are multiplied by the corresponding weight coefficients respectively, and then the sum is obtained to obtain the stain estimation TR of the corresponding textile;

Refine the defective area in the image. After the refinement is completed, use threshold segmentation to separate the defective area in the image from the background, and classify it according to the characteristics of the defective area, which include small hole characteristics and tear characteristics; set a weight coefficient corresponding to the small hole characteristics and the tear characteristics respectively; count the number of pixels in each defective area in the image, and convert it based on the resolution of the image to obtain each group of defect areas of the corresponding textile; accumulate the defect areas of the small hole characteristics as the small hole area of the corresponding textile; accumulate the defect areas of the tear characteristics as the tear area of the corresponding textile; extract the weight coefficient of each feature, multiply the small hole area and the tear area by the corresponding weight coefficient respectively, and then sum them to obtain the defect estimation TE of the corresponding textile.

In some embodiments, the defect estimation value Defal of the corresponding textile is obtained, specifically:

Identify the edges of the stitches in the image, and after the identification is completed, obtain the straight line segments in the image, which are the stitches of the corresponding textile; after obtaining the straight line segments, track the detected straight line segments, and identify the anomalies in the straight line segments in the process of analyzing their paths and continuity, the anomalies include bending and breaking; calculate the bending length and breaking length of each straight line segment in the image, and accumulate each group of bending lengths and breaking lengths to obtain the total bending length and total breaking length of the corresponding textile; set a weight coefficient corresponding to the bending and breaking respectively; multiply the total bending length and total breaking length of the corresponding textile by the corresponding weight coefficient respectively, and then sum them to obtain the stitch estimation TC of the corresponding textile;

Get the target workshop to which the corresponding textile belongs, set the maximum allowable values of the stain estimation TR, defect estimation TE and stitch estimation TC of textiles in different target workshops; substitute the stain estimation TR, defect estimation TE and stitch estimation TC of the corresponding textile into the formula Perform weighted calculation to obtain the defect estimation value Defal of the corresponding textile; , as well as They represent the maximum permissible values of the stain estimate TR, defect estimate TE and seam estimate TC respectively; , as well as They are the influencing weight factors of stain estimation TR, defect estimation TE and seam estimation TC respectively.

In some embodiments, the specific process of obtaining the defect level based on the defect estimation Defal of the corresponding textile is as follows:

Reference thresholds of defect valuation Defal for different target workshops are preset. If the defect valuation Defal of the corresponding textile is greater than the corresponding preset reference threshold, the difference between the two is further calculated as the defect excess value of the corresponding textile, and each group of value ranges of the defect excess value is preset. Each group of value ranges is set to correspond to a defect level; the defect excess value of the corresponding textile is matched with each group of preset value ranges to obtain the defect level of the corresponding textile.

In some embodiments, a maintenance signaling corresponding to a specific type of abnormality is triggered, and corresponding steps are performed, specifically:

Extract the fault detection point of the specific abnormal type, draw a circle with the fault detection point as the center and the set distance as the radius, and select the maintenance personnel within the circle as the candidate to solve the maintenance signaling;

Send a position feedback signal to the mobile terminal of each candidate for solution. After each candidate for solution confirms the position feedback signal, the specific location of each candidate for solution is obtained. Based on the specific location of each candidate for solution, the distance from each candidate for solution to the fault detection point is calculated and recorded as Jr.

Further obtain the historical number of solutions of each candidate for resolution, extract the solution time used for each exception solution from the historical number of solutions of each candidate for resolution, mark the historical number of solutions of each candidate for resolution as Je, take the average solution time of each group of candidates for resolution as the average solution time Jf of each candidate for resolution.

In some embodiments, a maintenance signaling corresponding to a specific type of abnormality is triggered, and corresponding steps are performed, further:

The specific types of anomalies are numbered, and the number is represented by i, where i=1, 2 or p, and p is the total number of specific types of anomalies; the influence weight factors of the distance Jr, the number of historical solutions Je, and the average solution time Jf corresponding to different specific types of anomalies are set, and the influence weight factors of the distance Jr, the number of historical solutions Je, and the average solution time Jf corresponding to the specific types of anomalies are marked as , as well as ;

Based on the specific type of abnormality that currently triggers the maintenance signaling, the formula Perform weighted calculation to obtain the preferred value wefgh of each candidate solution within the circle; sort the preferred value wefgh of each candidate solution from large to small, and select the candidate solution with the largest preferred value wefgh as the processing personnel who triggers the maintenance signaling this time, and the historical number of solutions of the processing personnel is increased by one; send the specific type of the current abnormality and the corresponding fault detection point to the mobile terminal of the processing personnel.

In some embodiments, the anomaly detection module is also used to monitor and analyze the defect excess values of textiles on each branch conveyor belt in real time, and preset a threshold number of the defect excess values of textiles on each branch conveyor belt. If the number of defect excess values of textiles on the corresponding branch conveyor belt during a certain sorting operation reaches the preset threshold number, a production optimization signal is triggered to the mobile terminal of the manager.

Compared with the prior art, the present invention has the following beneficial effects:

The present invention reduces manual input errors and speeds up sorting by automatically identifying textile labels. At the same time, when sorting textiles to corresponding target workshops, the image information of the textiles is collected and analyzed, stains, defects and seams are comprehensively evaluated, and defect valuations are generated, thereby providing comprehensive data support for quality control of the textiles and improving the accuracy of quality evaluation. Further, a barcode including defect grade information is generated according to the defect valuation of the corresponding textile, and the barcode is attached to the corresponding textile. After the attachment is completed, the textile is continuously transported until it is transported to the target workshop of the corresponding textile, thereby solving the problem in the prior art that the quality of the production of the textile cannot be evaluated based on the quality requirements of the textile, and the evaluation results and the textiles cannot be packaged and sent to the corresponding sorting target workshop.

The present invention monitors the sorting process in real time, quickly identifies and responds to anomalies when they occur, and selects personnel with the largest preferred value to solve the anomalies through a series of steps, thereby improving the efficiency of exception handling.

BRIEF DESCRIPTION OF THE DRAWINGS

Further details, features and advantages of the present application are disclosed in the following description of exemplary embodiments in conjunction with the accompanying drawings, in which:

FIG1 is a block diagram of the principle of the present invention.

DETAILED DESCRIPTION

Several embodiments of the present application will be described in more detail below with reference to the accompanying drawings so that those skilled in the art can implement the present application. The present application can be embodied in many different forms and purposes and should not be limited to the embodiments described herein. These embodiments are provided to make the present application comprehensive and complete, and to fully convey the scope of the present application to those skilled in the art. The embodiments do not limit the present application.

Unless otherwise defined, all terms (including technical and scientific terms) used herein have the same meaning as commonly understood by those of ordinary skill in the art to which this application belongs. It will be further understood that terms such as those defined in commonly used dictionaries should be interpreted as having a meaning consistent with their meaning in the relevant art and/or the context of this specification, and will not be interpreted in an idealized or overly formal sense unless explicitly defined as such herein.

Please refer to FIG1 , which shows an intelligent automatic textile sorting system based on machine vision, including a classification decision module, an image analysis module, a quality assessment module, and an anomaly detection module;

The classification decision module is used for the textiles to move with the conveyor belt, and when passing through the RFID reading area, the RFID tag on each textile is scanned and identified by the RFID reader, and the information in the RFID tag of each textile is parsed after identification, and the target workshop of each textile is determined, and the execution device is controlled according to the determined target workshop to transfer the corresponding textile from the main conveyor belt to the branch conveyor belt of the target workshop; the execution device is such as a robotic arm or a start valve; the tag information includes but is not limited to the number, material, type, color, etc.;

It should be noted that if the RFID tag cannot be read or displays incorrect information, the corresponding textile will be directly transported to the abnormal processing workshop, where operators in the workshop will check the textiles and manually sort or correct the tag information.

The image analysis module collects image information of each piece of textile through the image acquisition device set on each branch conveyor belt during the conveyance of each piece of textile through the corresponding branch conveyor belt; the image acquisition device is such as a high-resolution industrial camera; the image information of each piece of textile is collected, each group of image information is pre-processed and analyzed, and the color and pattern information matching the label is extracted from the analysis results of each group of image information; the color and pattern features are extracted from the pre-processed image by using image processing technology; and the color verification information and pattern information of the corresponding textile are matched and verified with the color and pattern information in the label respectively. If the verification is passed, the image information of the corresponding textile is sent to the quality assessment module, otherwise the corresponding textile is transported to the abnormal processing workshop; the pre-processing includes but is not limited to graying, filtering and edge detection, etc.

It should be noted that, assuming that the RFID tag of a textile indicates that it should have a red and white striped pattern, when the textile passes through the branch conveyor, the industrial camera captures its image. First, the red and white stripes are identified in the image, and then the arrangement and width of these stripes are verified to be consistent with the label description. If the verification is successful, the image information of the corresponding textile is sent to the quality assessment module; if the verification fails, the corresponding textile is automatically transported to the abnormal processing workshop.

The quality assessment module receives image information of the corresponding textile, and performs analysis to obtain a defect estimation value Defal of the corresponding textile; generates a barcode including defect grade information based on the defect estimation value Defal of the corresponding textile, attaches the barcode to the corresponding textile, and continues to transport the textile after the attachment is completed until it is transported to a target workshop of the corresponding textile;

The defect estimation value Defal of the corresponding textile is obtained, specifically:

S1: First, enhance the image; for example, use a filter to highlight the edges of stains and spots; after enhancement, use the threshold segmentation technology to separate the stain area in the image from the background, and classify it according to the characteristics of the stain area, including oil stain characteristics, ink characteristics, and dye penetration characteristics; set a weight coefficient corresponding to the oil stain characteristics, ink characteristics, and dye penetration characteristics respectively; the technicians preset it according to the specific characteristics, and it can be adjusted later;

Count the number of pixels of each stain area in the image, and convert it based on the resolution of the image to obtain the stain area of each group of corresponding textiles;

The stain areas of the corresponding oil stain characteristics are accumulated to obtain the oil stain area of the corresponding textile;

The stain areas of the corresponding ink features are accumulated to obtain the ink area of the corresponding textile;

The stain areas of the corresponding dye penetration characteristics are accumulated as the penetration area of the corresponding textile;

Extract the weight coefficient of each feature, multiply the oil stain area, ink area and penetration area by the corresponding weight coefficient respectively, and then sum them up to obtain the stain estimation TR of the corresponding textile;

S2: Apply morphological operations to refine the defective areas in the image, such as dilation, erosion, opening and closing operations, to remove small noise points and fill small holes. After the refinement, use threshold segmentation to separate the defective areas in the image from the background, and classify them according to the features of the defective areas, including small hole features and tear features. Set a weight coefficient for each of the small hole features and tear features.

Count the number of pixels in each defect area in the image, and convert it based on the resolution of the image to obtain each group of defect areas of the corresponding textile;

The defect areas of the corresponding small hole features are accumulated as the small hole area of the corresponding textile;

Accumulate the defect areas of the corresponding tear features as the tear area of the corresponding textile;

Extract the weight coefficients of each feature, multiply the small hole area and the tear area by the corresponding weight coefficients respectively, and then sum them up to obtain the defect estimation TE of the corresponding textile;

S3: Use the Canny edge detector to identify the edges of the stitches in the image. After identification, apply the Hough transform to obtain the straight line segments in the image. The straight line segments correspond to the stitches of the textile. After obtaining the straight line segments, track the detected straight line segments and identify anomalies in the straight line segments in the process of analyzing their paths and continuity. The anomalies include bends and broken lines. The bends are identified by comparing the difference between the actual path of the straight line segment and the fitted straight line.

Calculate the bending length and the broken line length of each straight line segment in the image, and accumulate each group of bending length and broken line length to obtain the total bending length and total broken line length of the corresponding textile;

A weight coefficient is set to correspond to bending and breaking respectively; the total bending length and the total breaking length of the corresponding textile are multiplied by the corresponding weight coefficient respectively, and then the sum is obtained to obtain the estimated stitching TC of the corresponding textile;

S4: Obtain the target workshop to which the corresponding textile belongs, and set the maximum allowable values of the stain estimation TR, defect estimation TE, and stitch estimation TC of textiles in different target workshops; the technical personnel shall set the values according to the application scenarios and types of textiles in different target workshops;

Substitute the stain estimate TR, defect estimate TE and stitch estimate TC of the corresponding textile into the formula Perform weighted calculation to obtain the defect estimation value Defal of the corresponding textile; , as well as They represent the maximum permissible values of the stain estimate TR, defect estimate TE and seam estimate TC respectively; , as well as are the influencing weight factors of stain estimation TR, defect estimation TE and seam estimation TC, and their values are set to 1.205, 1.218 and 1.209 respectively;

S5: Preset reference thresholds of defect estimation values Defal for different target workshops. If the defect estimation value Defal of the corresponding textile is greater than the corresponding preset reference threshold, further calculate the difference between the two as the defect excess value of the corresponding textile, preset each group of value ranges of the defect excess value, and set each group of value ranges to correspond to a defect level; match the defect excess value of the corresponding textile with each group of preset value ranges to obtain the defect level of the corresponding textile; the larger the defect excess value, the higher the corresponding matching defect level;

It should be noted that, in summary, quality assessment is achieved while sorting textiles, which saves labor costs and reduces possible errors in manual sorting and quality inspection. The quality of textiles is comprehensively evaluated based on the calculated defect valuation, providing data reference for subsequent adjustments to the textile production process.

The abnormality detection module is used to monitor the sorting process in real time, and identify the specific type of abnormality through the set self-check mechanism when an abnormality occurs; the specific types of abnormalities include but are not limited to label reading failure, image acquisition failure, classification error or actuator failure, etc. After the identification is completed, the maintenance signaling corresponding to the specific type of abnormality is triggered, and the corresponding steps are executed;

Trigger the maintenance signaling corresponding to the specific type of exception and execute the corresponding steps, specifically:

Extract the fault detection point of the specific type of abnormality; for example, if the image acquisition fails, the image acquisition device is used as the fault detection point, and if the tag reading fails, the RFID reader is used as the fault detection point. The specific fault detection point is set by the technician and can be adjusted according to the actual application situation later; draw a circle with the fault detection point as the center and the set distance as the radius, and select the maintenance personnel within the circle as the candidates for maintenance signaling resolution;

Send a position feedback signal to the mobile terminal of each candidate for solution. After each candidate for solution confirms the position feedback signal, the specific location of each candidate for solution is obtained. Based on the specific location of each candidate for solution, the distance from each candidate for solution to the fault detection point is calculated and recorded as Jr.

Further obtain the historical resolution times of each candidate solver, and extract the resolution time used for each exception resolution from the historical resolution times of each candidate solver; the resolution time starts from the time when the solver arrives at the fault detection point and ends when the exception resolution is completed;

Mark the historical number of solutions of each candidate as Je; take the average of the solution time of each group of candidates as the average solution time Jf of each candidate;

Number the specific types of exceptions, and the number is represented by i, where i=1, 2 or p, where p is the total number of specific types of exceptions;

Set different influence weight factors of the distance Jr, historical resolution times Je and average resolution time Jf corresponding to different specific types of anomalies, and mark the influence weight factors of the distance Jr, historical resolution times Je and average resolution time Jf corresponding to the specific types of anomalies as , as well as ;

Based on the specific type of abnormality that currently triggers the maintenance signaling, the formula Perform weighted calculation to obtain the optimal value wefgh of each candidate within the circle; , as well as Specific values are taken based on the specific type of the current exception;

Based on the preferred value wefgh of each candidate solution, sort them from large to small, select the candidate solution with the largest preferred value wefgh as the processing personnel who triggers the maintenance signaling, and the historical number of processing personnel is increased by one;

Send the specific type of the current abnormality and the corresponding fault detection point to the mobile terminal of the processing personnel;

It should be noted that different impact weight factors are set according to different abnormality types so that it can flexibly adapt to different maintenance scenarios. For example, tag reading failure is relatively easy to solve, so the weight of priority distance is higher. Actuator failure means that the abnormality problem is more serious and requires more experienced and efficient personnel to handle it, so the weight of average solution time and historical solution times is higher.

It should be noted that, in summary, the sorting process is monitored, and when an abnormality occurs, it is quickly identified and responded to. Through a series of steps, the personnel with the largest wefgh value are selected to solve the abnormality, thereby improving the processing efficiency.

The anomaly detection module is also used to monitor and analyze the defect excess value of textiles on each branch conveyor belt in real time, and preset the threshold number of the defect excess value of textiles on each branch conveyor belt. If the number of the defect excess value of textiles on the corresponding branch conveyor belt during a sorting operation reaches the preset threshold number, the production optimization signal is triggered to the mobile terminal of the manager;

It should be noted that the above-mentioned system monitors and analyzes the defect excess values of textiles on each branch conveyor belt in real time, and presets the corresponding threshold number. When the preset threshold number is reached, it means that the number of defective textiles is large, then the corresponding production optimization signal is triggered to remind the management personnel to adjust and optimize the problems existing in the textile production line, thereby further improving the degree of intelligence.

The above formulas are obtained by collecting a large amount of data and performing software simulation, and a formula close to the actual value is selected. The influencing weight factor and specific coefficient value in the formula are set by technical personnel in this field according to actual conditions, and can be adjusted and modified later.

The preferred embodiments of the present invention disclosed above are only used to help explain the present invention. The preferred embodiments do not describe all the details in detail, nor do they limit the invention to only specific implementation methods. Obviously, many modifications and changes can be made according to the content of this specification. This specification selects and specifically describes these embodiments in order to better explain the principles and practical applications of the present invention, so that those skilled in the art can understand and use the present invention well. The present invention is limited only by the claims and their full scope and equivalents.

Claims (4)

1. An intelligent textile automatic sorting system based on machine vision, characterized by comprising: Classification decision module: When the textile moves along the conveyor belt and passes through the RFID reading area, the RFID tag on each textile is scanned and identified. After identification, the information in the RFID tag of each textile is parsed to determine the target workshop for each textile. According to the determined target workshop, the execution device is controlled to transfer the corresponding textile from the main conveyor belt to the branch conveyor belt of the target workshop. Image analysis module: when each piece of textile is conveyed through the corresponding branch conveyor belt, the image acquisition device set on each branch conveyor belt collects the image information of each piece of textile, pre-processes and analyzes each group of image information, extracts the color and pattern information matching the label from the analysis results of each group of image information, and matches and verifies the color verification information and pattern information of the corresponding textile with the color and pattern information in the label respectively. If the verification is passed, the image information of the corresponding textile is sent to the quality assessment module; Quality assessment module: receiving image information of the corresponding textile, and analyzing it to obtain a defect estimation value Defal of the corresponding textile; generating a barcode including defect grade information based on the defect estimation value Defal of the corresponding textile, attaching the barcode to the corresponding textile, and continuing to transport it after the attachment is completed until it is transported to the target workshop of the corresponding textile; Specifically: First, the image is enhanced, and after the enhancement, the stain area in the image is separated from the background, and classified according to the characteristics of the stain area, including oil stain characteristics, ink characteristics and dye penetration characteristics, and a weight coefficient is set for the oil stain characteristics, ink characteristics and dye penetration characteristics respectively; the number of pixels of each stain area in the image is counted, and the image is converted based on the resolution to obtain each group of stain areas of the corresponding textile; the stain areas of the oil stain characteristics are accumulated as the oil stain area of the corresponding textile; the stain areas of the ink characteristics are accumulated as the ink area of the corresponding textile; the stain areas of the dye penetration characteristics are accumulated as the penetration area of the corresponding textile; the weight coefficients of each feature are extracted, the oil stain area, ink area and penetration area are multiplied by the corresponding weight coefficients respectively, and then the sum is obtained to obtain the stain estimation TR of the corresponding textile; Refine the defective area in the image, separate the defective area in the image from the background by threshold segmentation after refinement, and classify according to the features of the defective area, which include small hole features and tear features; set a weight coefficient corresponding to the small hole features and tear features respectively; count the number of pixels in each defective area in the image, and convert based on the resolution of the image to obtain each group of defect areas of the corresponding textile; accumulate the defect areas of the small hole features as the small hole area of the corresponding textile; accumulate the defect areas of the tear features as the tear area of the corresponding textile; extract the weight coefficients of each feature, multiply the small hole area and the tear area by the corresponding weight coefficients respectively, and then sum them to obtain the defect estimation TE of the corresponding textile; Identify the edges of the stitches in the image, and after the identification is completed, obtain the straight line segments in the image, which are the stitches of the corresponding textile; after obtaining the straight line segments, track the detected straight line segments, and identify the anomalies in the straight line segments in the process of analyzing their paths and continuity, the anomalies include bending and breaking; calculate the bending length and breaking length of each straight line segment in the image, and accumulate each group of bending lengths and breaking lengths to obtain the total bending length and total breaking length of the corresponding textile; set a weight coefficient corresponding to the bending and breaking respectively; multiply the total bending length and total breaking length of the corresponding textile by the corresponding weight coefficient respectively, and then sum them to obtain the stitch estimation TC of the corresponding textile; Get the target workshop to which the corresponding textile belongs, set the maximum allowable values of the stain estimation TR, defect estimation TE and stitch estimation TC of textiles in different target workshops; substitute the stain estimation TR, defect estimation TE and stitch estimation TC of the corresponding textile into the formula A weighted calculation is performed to obtain the defect estimation Defal of the corresponding textile; wherein TR _allowed , TE _allowed and TC _allowed represent the maximum allowable values of the stain estimation TR, the defect estimation TE and the stitch estimation TC respectively; α1, α2 and α3 are the influencing weight factors of the stain estimation TR, the defect estimation TE and the stitch estimation TC respectively; Preset reference thresholds of defect estimation values Defal for different target workshops; if the defect estimation value Defal of the corresponding textile is greater than the corresponding preset reference threshold, further calculate the difference between the two as the defect excess value of the corresponding textile; preset each group of value ranges of the defect excess value, and set each group of value ranges to correspond to a defect grade; match the defect excess value of the corresponding textile with each group of preset value ranges to obtain the defect grade of the corresponding textile; Abnormality detection module: monitors the sorting process in real time, and identifies the specific type of abnormality through the set self-check mechanism when an abnormality occurs. After the identification is completed, the maintenance signal corresponding to the specific type of abnormality is triggered, and the corresponding steps are executed. 2. According to claim 1, a machine vision-based intelligent textile automatic sorting system is characterized in that the maintenance signaling corresponding to the specific type of abnormality is triggered and the corresponding steps are executed, specifically: Extract the fault detection point of the specific abnormal type, draw a circle with the fault detection point as the center and the set distance as the radius, and select the maintenance personnel within the circle as the candidate to solve the maintenance signaling; Send a position feedback signal to the mobile terminal of each candidate for solution. After each candidate for solution confirms the position feedback signal, the specific location of each candidate for solution is obtained. Based on the specific location of each candidate for solution, the distance from each candidate for solution to the fault detection point is calculated and recorded as Jr. Further obtain the historical number of solutions of each candidate for resolution, extract the solution time used for each exception solution from the historical number of solutions of each candidate for resolution, mark the historical number of solutions of each candidate for resolution as Je, take the average solution time of each group of candidates for resolution as the average solution time Jf of each candidate for resolution. 3. According to claim 2, a machine vision-based intelligent textile automatic sorting system is characterized in that the maintenance signaling corresponding to the specific type of abnormality is triggered and the corresponding steps are executed, further: The specific types of anomalies are numbered, and the number is represented by i, where i = 1, 2 or p, and p is the total number of specific types of anomalies; the influence weight factors of the distance Jr, the number of historical solutions Je, and the average solution time Jf corresponding to different specific types of anomalies are set, and the influence weight factors of the distance Jr, the number of historical solutions Je, and the average solution time Jf corresponding to the specific types of anomalies are marked as ηi, μi, and λi respectively; Based on the specific type of abnormality that currently triggers the maintenance signaling, the formula Perform weighted calculation to obtain the preferred value wefgh of each candidate solution within the circle; sort the candidate solution from large to small based on the preferred value wefgh of each candidate solution, and select the candidate solution with the largest preferred value wefgh as the processing personnel who triggers the maintenance signaling this time, and the historical number of solutions of the processing personnel is increased by one; send the specific type of the current abnormality and the corresponding fault detection point to the mobile terminal of the processing personnel. 4. According to claim 1, an intelligent textile automatic sorting system based on machine vision is characterized in that the anomaly detection module is also used to monitor and analyze the defect excess value of textiles on each branch conveyor belt in real time, and a threshold number of textile defect excess values on each branch conveyor belt is preset. If the number of textile defect excess values on the corresponding branch conveyor belt during a certain sorting operation reaches the preset threshold number, a production optimization signal is triggered to the mobile terminal of the manager.