Cloth surface defect detection method and system based on machine vision

Abstract

The invention relates to the technical field of cloth surface defect detection, and particularly discloses a cloth surface defect detection method and system based on machine vision, and the method comprises the steps: analyzing the pilling characteristics of a surface image of cloth, obtaining the number ratio of abnormal regions, judging the qualification of the cloth, and generating an unqualified signal and a suspected qualified signal; analyzing the abnormal marking result of the cloth surface area to obtain a pilling trend value, determining the influence of the cloth surface pilling on the cloth surface defect, and generating a strong influence signal; according to the method, the fabric detection time is optimized by determining the fabric optimal detection time and the detection time of the fabric production equipment and analyzing the fabric production detection time, the detection time of the fabric is optimized, the detection time of the fabric production equipment is optimized, the detection time of the fabric production equipment is optimized, and the detection accuracy of the fabric production equipment is improved. The detection pertinence and efficiency are improved, and problems in the production process can be found and solved in time, so that the quality of cloth is ensured.

Description

Technical Field

[0001] The present invention relates to the technical field of cloth surface defect detection, and in particular to a cloth surface defect detection method and system based on machine vision. Background Art

[0002] With the rapid development of computer technology and artificial intelligence technology, machine vision technology has gradually become an important tool in the field of automated inspection and quality control. Machine vision technology can recognize, analyze and process images and videos by simulating human visual functions. The textile industry is an important part of the national economy, providing people with clothing, curtains, bedding and other items that are indispensable in daily life. In order to ensure the quality of textiles, textile manufacturers need to conduct strict quality inspections on the surface of the fabrics. However, existing methods for detecting surface defects of fabrics often rely on manual visual inspection after fabric production is completed. However, this method is inefficient and the detection results are easily affected by human factors. It may lead to the omission of defective fabrics, thus affecting the quality of fabric shipment.

[0003] To this end, we propose a cloth surface defect detection method and system based on machine vision. Summary of the invention

[0004] The purpose of the present invention is to provide a method and system for detecting cloth surface defects based on machine vision to solve the technical problems in the above background.

[0005] The purpose of the present invention can be achieved through the following technical solutions: A method for detecting surface defects of cloth based on machine vision, comprising: Analyze the pilling characteristics of the surface image of the fabric, obtain the number ratio of abnormal areas, and determine whether the fabric generates an unqualified signal or a suspected qualified signal; Based on the suspected qualified signal, the abnormal marking results of the fabric surface area are analyzed to obtain the pilling trend value, determine the impact of fabric surface pilling on fabric surface defects, and generate a strong impact signal; Extract all fabric samples with unqualified signals and strong impact signals within the test time, record them as unqualified traceability samples and strong impact traceability samples respectively, obtain the production time, analyze them, and output the regular data of the traceability samples; The time of adjacent inspections for unqualified conditions and strong impact conditions during fabric production is analyzed to obtain the inspection time of fabric production equipment.

[0006] As a further technical solution of the present invention: the process of judging whether the fabric generates an unqualified signal or a suspected qualified signal is as follows: Identify whether pilling exists in all areas of the fabric surface image. If so, the corresponding area is recorded as an abnormal area. If not, the corresponding area is recorded as a normal area. Calculate the number ratio of abnormal areas. If the number ratio is ≤ the abnormal area ratio threshold, generate a suspected qualified signal. If the number ratio is > the abnormal area ratio threshold, generate an unqualified signal.

[0007] As a further technical solution of the present invention: the process of obtaining the pilling tendency value is as follows: Analyze the pilling area of ​​the abnormal area and conduct pilling test on the normal area to obtain the pilling performance value of the abnormal area and the increase coefficient of the normal area; The pilling trend value is obtained by performing ratio processing on the pilling characteristic value of the abnormal area and the increase and change coefficient of the normal area.

[0008] As a further technical solution of the present invention: the process of obtaining the pilling performance value of the abnormal area is: Extract the pilling area of ​​the abnormal area, and if it is greater than or equal to the standard value of the pilling area, mark the corresponding pilling as a prominent pilling; Obtaining the average pilling area of ​​all prominent pillings, and calculating the deviation ratio from the standard value of the pilling area to obtain the prominent pilling area deviation ratio; The ratio of the number of prominent pilling and the deviation ratio of the prominent pilling area are summed to obtain the pilling performance value of the abnormal area.

[0009] As a further technical solution of the present invention: the process of obtaining the mutation coefficient of the normal area is: The pilling test is carried out on the normal area to obtain the pilling grade parameters, which are then compared with the standard pilling grade parameters to obtain the increase coefficient of the normal area.

[0010] As a further technical solution of the present invention: the process of outputting the regular data of the traceable sample includes: Obtain the production time of the unqualified traceability sample and the production time of the strong-impact traceability sample, and divide them into the production time interval of the unqualified traceability sample and the production time interval of the strong-impact traceability sample; Subtract the production time intervals of adjacent unqualified traceability samples to obtain the interval intervals of unqualified traceability samples, calculate the variance of the intervals of all unqualified traceability samples, and obtain the time variance value of strongly influential traceability samples; If it is greater than the time variance threshold of the unqualified traceability sample, an unqualified non-periodic signal is generated; If it is less than or equal to the time variance threshold of the unqualified traceability sample, an unqualified periodic signal is generated.

[0011] As a further technical solution of the present invention: the process of outputting the regular data of the traceable sample also includes: Subtract the production time intervals of adjacent strong-impact traceability samples to obtain the interval intervals of the strong-impact traceability samples, calculate the variance of the intervals of all strong-impact traceability samples, and obtain the time variance value of the strong-impact traceability samples; If it is greater than the time variance threshold of the strong impact tracing sample, a strong impact aperiodic signal is generated; If it is less than or equal to the time variance threshold of the strong impact tracing sample, a strong impact periodic signal is generated.

[0012] As a further technical solution of the present invention: the process of obtaining the time of adjacent detections is: Get the mean of the intervals of all unqualified traceability samples, which is the time of adjacent tests for unqualified cases during the production of the fabric; The mean of the intervals of all strong-impact traceability samples is obtained, which is the time of adjacent inspections of the fabric for strong-impact situations during production.

[0013] As a further technical solution of the present invention: the process of obtaining the detection time of the cloth production equipment is: If the intervals of all strongly-influenced traceability samples are located before the intervals of unqualified traceability samples in terms of time, and the intervals of both are smaller than the preset intervals, the start time of adjacent tests of the strongly-influenced situation shall be recorded as the test time of the fabric production equipment.

[0014] A cloth surface defect detection system based on machine vision, the system comprising: Preliminary detection module: Analyze the pilling characteristics of the surface image of the fabric, obtain the number ratio of abnormal areas, and determine whether the fabric generates an unqualified signal or a suspected qualified signal; In-depth detection module: Based on the suspected qualified signal, analyze the abnormal marking results of the fabric surface area, obtain the pilling trend value, determine the impact of fabric surface pilling on fabric surface defects, and generate a strong impact signal; Periodicity judgment module: extract all fabric samples with unqualified signals and strong impact signals within the test time, record them as unqualified traceability samples and strong impact traceability samples respectively, obtain the production time, analyze them, and output the regular data of the traceability samples; Optimal detection time determination module: Analyzes the time of adjacent detections for unqualified conditions and strong impact conditions during fabric production to obtain the detection time of fabric production equipment.

[0015] Beneficial effects of the present invention: (1) The present invention analyzes the pilling characteristics of the surface image of the fabric to obtain the proportion of abnormal areas, judge the quality of the fabric, and generate unqualified signals and suspected qualified signals; based on the suspected qualified signals, the abnormal marking results of the fabric surface area are analyzed to obtain the pilling trend value, determine the influence of the pilling on the fabric surface defects, and generate a strong influence signal. The present invention combines image processing technology and data analysis methods to realize the detection and evaluation of pilling on the fabric surface and improve the detection accuracy.

[0016] (2) The present invention extracts all fabric samples with unqualified signals and strong impact signals within the test time, records them as unqualified traceability samples and strong impact traceability samples respectively, obtains the production time, performs analysis, and outputs the regular data of the traceability samples; analyzes the time of adjacent detections for unqualified conditions and strong impact conditions during fabric production to obtain the detection time of the fabric production equipment; the present invention determines the fabric optimization detection time and the detection time of the fabric production equipment, which is convenient for adjusting the detection time according to the quality change of the fabric produced by the current equipment, effectively reducing the current frequency of fabric detection, and using the start time of adjacent detection of strong impact conditions to determine the detection time of the fabric equipment during production, so that under this time, the fabric equipment can be overhauled in advance, which can effectively improve the quality of subsequent fabric production. Therefore, the analysis of the fabric production detection time optimizes the fabric detection time, which not only improves the pertinence and efficiency of the detection, but also helps to timely discover and solve problems in the production process, thereby ensuring the quality of the fabric. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] The present invention will be further described below in conjunction with the accompanying drawings.

[0018] Figure 1 It is a flowchart of a method for detecting surface defects of cloth based on machine vision according to an embodiment of the present invention; Figure 2 The system block diagram of a cloth surface defect detection system based on machine vision according to an embodiment of the present invention. DETAILED DESCRIPTION

[0019] 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.

[0020] Example 1 See also Figure 1As shown, a method for detecting surface defects of cloth based on machine vision according to an embodiment of the present invention is specifically described as follows: Step 1: Analyze the pilling characteristics of the surface image of the fabric, obtain the percentage of abnormal areas, judge the quality of the fabric, and generate unqualified signals and suspected qualified signals; In some embodiments, a test time is set, and all current test fabric samples are analyzed, wherein the test time corresponds to the production time of the fabric samples, and the samples are arranged in ascending order: a surface image of the fabric is acquired using a scanner, and the surface image of the fabric is divided into a plurality of regions in a grid form, and pilling features of the fabric surface are extracted according to an edge detection algorithm, specifically: the fabric surface image is subjected to image preprocessing, an edge detection operator (such as a Canny operator, a Sobel operator, a Robert operator, a Prewitt operator) is used to identify the pilling region by detecting edge information in the image, and morphological processing (such as dilation, erosion, opening operation, closing operation) is performed to extract the pilling features of the pilling region; Identify whether there is pilling in the surface area of ​​the fabric. If so, obtain the area corresponding to the pilling distribution position and mark it as an abnormal area. Mark the area without pilling in the surface area of ​​the fabric as a normal area. Count the number of abnormal areas and perform ratio processing on it with the total number of divided areas to obtain the ratio of the number of abnormal areas. Compare the percentage of abnormal areas with the threshold of the percentage of abnormal areas: If the percentage of abnormal regions is greater than the threshold of the percentage of abnormal regions, it means that pilling is more distributed on the surface of the fabric, and a failure signal is generated; If the percentage of abnormal regions is less than or equal to the threshold of the percentage of abnormal regions, it means that the pilling is less distributed on the surface of the fabric, and a suspected qualified signal is generated; For example, the surface image of the fabric is divided into 50*50 rectangular areas. If 80 abnormal areas with pilling are identified in the surface image of the fabric, the proportion of the number of abnormal areas in the surface image of the fabric is 80 / (50*50)=0.032; Through the above analysis, the abnormal area can be determined, and then the pilling defect position can be determined in the process of fabric surface defect detection, and the degree of the defect can be determined according to the proportion of the number of abnormal areas; therefore, step 1 of the present invention can realize the positioning of the defect position and the judgment of the degree of the defect; Step 2: Based on the suspected qualified signal, analyze the abnormal marking results of the fabric surface area, obtain the pilling trend value, determine the impact of fabric surface pilling on fabric surface defects, and generate a strong impact signal; Compare the pilling area in the abnormal area when the suspected pass signal is generated with the pilling area standard value: If the pilling area is greater than or equal to the standard value of the pilling area, the corresponding pilling will be marked as a prominent pilling; If the pilling area is smaller than the standard value of the pilling area, the corresponding pilling will be marked as non-protruding pilling; It should be noted that the standard value of pilling area is set by those skilled in the art based on fabric design requirements; Count the number of prominent pilling and the number of non-prominent pilling, and sum them up to get the total number of pilling, and perform ratio processing on the number of prominent pilling and the total number of pilling to get the proportion of prominent pilling; Obtaining the pilling area in the prominent pilling, performing a subtraction process on the pilling area and the standard value of the pilling area to obtain the pilling area deviation, summing up all the pilling area deviations and taking the average value to obtain the mean value of the pilling area deviation, performing a ratio process on the mean value of the pilling area deviation and the standard value of the pilling area to obtain the prominent pilling area deviation ratio; The percentage of prominent pilling and the deviation ratio of prominent pilling area are summed to obtain the pilling performance value of the abnormal area. It should be noted that the pilling performance value is obtained by processing the percentage of prominent pilling numbers and the deviation ratio of prominent pilling area. The percentage of prominent pilling numbers reflects the number of large pilling areas in the abnormal area when the suspected qualified signal is generated. The deviation ratio of prominent pilling area reflects the degree of deviation between the prominent pilling area and the standard value of the pilling area in the abnormal area when the suspected qualified signal is generated. The larger the deviation, the more serious the pilling trend in the abnormal area, indicating that the greater the surface defect of the fabric; Get a normal area of ​​the fabric surface area to test for pilling; Specifically, the pilling test time is preset, the fabric is loaded on the sample holder, and the fabric sample is first rubbed with nylon and then with the standard fabric in a circular motion under a certain pressure to cause pilling. After a certain number of times, the fabric sample is compared with the original fabric or the standard sample in the rating box to obtain the pilling grade parameters, and the pilling grade parameters are compared with the standard pilling grade parameters to obtain the increase coefficient of the normal area; It should be noted that the standard pilling grade parameter is set by technical personnel in this field according to actual needs. The larger the pilling grade parameter, the worse the pilling effect and the better the anti-pilling effect. The maximum is 5 and the minimum is 1. 5 indicates mild pilling and 1 indicates severe pilling. When the result is between two adjacent grades, 0.5 is taken. The smaller the pilling grade parameter, the smaller the increase coefficient and the more severe the pilling trend. The pilling tendency value is obtained by performing ratio processing on the pilling characteristic value of the abnormal area and the increase coefficient of the normal area; Compare the Pilling Tendency value to the Pilling Tendency Threshold: If the pilling trend value is greater than or equal to the pilling trend threshold, it means that pilling has a greater impact on the surface defects of the fabric and a strong impact signal is generated; If the pilling trend value is less than the pilling trend threshold, it means that pilling has little impact on the surface defects of the fabric, and a weak impact signal is generated; It should be noted that the unqualified signals, strong influence signals and weak influence signals obtained by the fabric sample detection reflect the degree of defects of the fabric samples. Specifically, the fabric samples with unqualified signals have the greatest degree of defects, the fabric samples with strong influence signals have the second greatest degree of defects, and the fabric samples with weak influence signals have the smallest degree of defects. In turn, the degree of influence on the fabric samples in the later use is opposite. The fabric samples with unqualified signals have the worst use effect, the fabric samples with strong influence signals have the second greatest use effect, and the fabric samples with weak influence signals have very little use effect. The technical solution of the embodiment of the present invention is mainly as follows: analyzing the pilling characteristics of the surface image of the fabric, obtaining the proportion of the number of abnormal areas, judging the fabric qualification, and generating an unqualified signal and a suspected qualified signal; based on the suspected qualified signal, analyzing the abnormal marking results of the fabric surface area, obtaining the pilling trend value, determining the influence of the fabric surface pilling on the fabric surface defects, and generating a strong influence signal. The present invention realizes the detection and evaluation of pilling on the fabric surface by combining image processing technology and data analysis methods, thereby improving the detection accuracy.

[0021] Example 2 Based on Example 1, please refer to Figure 1 As shown, a method for detecting surface defects of cloth based on machine vision according to an embodiment of the present invention is specifically described as follows: Step 3: Extract all fabric samples with unqualified signals and strong impact signals within the test time, record them as unqualified traceability samples and strong impact traceability samples respectively, obtain the production time, analyze them, and output the regular data of the traceability samples; In some embodiments, all fabric samples with unqualified signals and strong impact signals within the test time are obtained, and are marked as unqualified traceability samples and strong impact traceability samples according to different signals, and the production time of the unqualified traceability samples and the production time of the strong impact traceability samples are obtained through a timer; According to time continuity (time continuity refers to fabric samples with the same signal appearing at adjacent times), the production time of unqualified traceability samples and the production time of strongly influential traceability samples are divided respectively, and the production time intervals of multiple unqualified traceability samples and the production time intervals of strongly influential traceability samples are obtained; Analyze the unqualified traceability samples, make a difference between the production time intervals of adjacent unqualified traceability samples, calculate the interval intervals of the unqualified traceability samples, and then integrate the interval intervals of all unqualified traceability samples in order of production time from small to large to construct an interval interval data group of unqualified traceability samples, calculate the variance of the data group, and obtain the time variance value of the unqualified traceability samples; Compare the failed traceability sample time variance value with the failed traceability sample time variance threshold: If the time variance value of the unqualified traceability sample is greater than the time variance threshold of the unqualified traceability sample, an unqualified non-periodic signal is generated; If the time variance value of the unqualified traceability sample is less than or equal to the time variance threshold of the unqualified traceability sample, an unqualified periodic signal is generated; Similarly, for the strong impact traceability samples, the production time intervals of adjacent strong impact traceability samples are subtracted to calculate the interval intervals of the strong impact traceability samples, and then all the interval intervals of the strong impact traceability samples are integrated in order of production time from small to large to construct the interval interval data group of the strong impact traceability samples, and the variance of the data group is calculated to obtain the time variance value of the strong impact traceability samples; Compare the strong impact tracing sample time variance value with the strong impact tracing sample time variance threshold: If the time variance value of the strong impact tracing sample is greater than the time variance threshold of the strong impact tracing sample, a strong impact aperiodic signal is generated; If the time variance value of the strong impact tracing sample is less than or equal to the time variance threshold of the strong impact tracing sample, a strong impact periodic signal is generated; It should be noted that: when a periodic signal is obtained, it can be effectively explained that within the time of the traced sample, the equipment used for fabric processing may have intermittent periodic faults, such as periodic abnormalities in bearings, belts, engines or any other components in the production line of the fabric processing equipment, thereby affecting the quality of the fabric products. Therefore, step 4 of the present invention not only makes a regular judgment on the quality of the fabric products, but also deduces the nature of the fault of the fabric processing equipment based on the generated signal; Step 4: Analyze the time of adjacent inspections for unqualified conditions and strong impact conditions during fabric production to obtain the inspection time of fabric production equipment; In some embodiments, if the periodic signal of the unqualified sample and the periodic signal of the strongly affected sample are generated at the same time, the optimal detection time of the fabric is analyzed, and the specific process is as follows: First, the intervals of the unqualified traceability samples are obtained, and the intervals of all the unqualified traceability samples are averaged to obtain the average intervals of the unqualified traceability samples. According to the average intervals of the unqualified traceability samples, the time of adjacent tests for unqualified conditions during the production of the fabric is determined; and according to the number of intervals of the unqualified traceability samples, the frequency of testing for unqualified conditions during the production of the fabric is determined; Then, the intervals of the strong impact traceability samples are obtained, and the intervals of all the strong impact traceability samples are averaged to obtain the average intervals of the strong impact traceability samples. According to the average intervals of the strong impact traceability samples, the time of adjacent tests for the strong impact situation during the production of the fabric is determined; and according to the number of intervals of the strong impact traceability samples, the detection frequency of the fabric for the strong impact situation during the production is determined; Furthermore, the intervals of the unqualified traceability samples and the intervals of the strong-impact traceability samples are compared one by one in order of time from small to large. If the intervals of all the strong-impact traceability samples are located before the intervals of the unqualified traceability samples in terms of time, and the intervals of both are smaller than the preset intervals, then based on the change in the degree of fabric defects during the production time, the fabric defects will increase with the extension of time within the time interval, and the start time of the adjacent detection of the strong-impact situation is recorded as the detection time of the fabric production equipment; Therefore, by determining the optimal detection time of the fabric and the detection time of the fabric production equipment, it is convenient to make adaptability adjustments to the detection time for the fabric produced by the current equipment according to its quality changes, effectively reducing the current frequency of fabric detection, and using the start time of adjacent detections with strong influence to determine the detection time during fabric equipment production, so that at this time, the fabric equipment can be overhauled in advance, which can effectively improve the quality of subsequent fabric production.

[0022] The technical solution of the embodiment of the present invention is mainly as follows: extracting all fabric samples with unqualified signals and strong impact signals within the test time, recording them as unqualified traceability samples and strong impact traceability samples respectively, and obtaining the production time, analyzing them, and outputting the regular data of the traceability samples; analyzing the time of adjacent detections of unqualified conditions and strong impact conditions during the production of fabrics, and obtaining the detection time of the fabric production equipment; the present invention determines the fabric preferred detection time and the detection time of the fabric production equipment, which is convenient for the fabric produced by the current equipment, according to its quality change, to make an adaptability adjustment to the detection time, effectively reducing the current frequency of fabric detection, and using the start time of adjacent detections of strong impact conditions to determine the detection time of the fabric equipment during production, so that at this time, the fabric equipment is overhauled in advance, which can effectively improve the quality of subsequent fabric production. Therefore, the analysis of the fabric production detection time optimizes the fabric detection time, which not only improves the pertinence and efficiency of the detection, but also helps to timely discover and solve problems in the production process, thereby ensuring the quality of the fabric.

[0023] Example 3 Based on Example 1 and Example 2, please refer to Figure 2 As shown, a cloth surface defect detection system based on machine vision according to an embodiment of the present invention includes: Preliminary detection module: Analyze the pilling characteristics of the surface image of the fabric, obtain the percentage of abnormal areas, judge the quality of the fabric, and generate unqualified signals and suspected qualified signals; In-depth detection module: Based on the suspected qualified signal, analyze the abnormal marking results of the fabric surface area, obtain the pilling trend value, determine the impact of fabric surface pilling on fabric surface defects, and generate a strong impact signal; Periodicity judgment module: extract all fabric samples with unqualified signals and strong impact signals within the test time, record them as unqualified traceability samples and strong impact traceability samples respectively, obtain the production time, analyze them, and output the regular data of the traceability samples; Optimal detection time determination module: Analyzes the time of adjacent detections for unqualified conditions and strong impact conditions during fabric production to obtain the detection time of fabric production equipment.

[0024] The above shows and describes the basic principles, main features and advantages of the present invention. It should be understood by those skilled in the art that the present invention is not limited to the above embodiments. The above embodiments and descriptions are only for explaining the principles of the present invention. Without departing from the spirit and scope of the present invention, the present invention may have various changes and improvements, which fall within the scope of the present invention. The scope of protection of the present invention is defined by the attached claims and their equivalents.

Claims

1. A method for detecting surface defects of cloth based on machine vision, characterized in that: include: Analyze the pilling characteristics of the surface image of the fabric, obtain the number ratio of abnormal areas, and determine whether the fabric generates an unqualified signal or a suspected qualified signal; Based on the suspected qualified signal, the abnormal marking results of the fabric surface area are analyzed to obtain the pilling trend value, determine the impact of fabric surface pilling on fabric surface defects, and generate a strong impact signal; Extract all fabric samples with unqualified signals and strong impact signals within the test time, record them as unqualified traceability samples and strong impact traceability samples respectively, obtain the production time, analyze them, and output the regular data of the traceability samples; The time of adjacent inspections for unqualified conditions and strong impact conditions during fabric production is analyzed to obtain the inspection time of fabric production equipment.

2. The method for detecting surface defects of fabrics based on machine vision according to claim 1, characterized in that: The process of determining whether the fabric generates an unqualified signal or a suspected qualified signal is as follows: Identify whether pilling exists in all areas of the fabric surface image. If so, the corresponding area is recorded as an abnormal area. If not, the corresponding area is recorded as a normal area. Calculate the number ratio of abnormal areas. If the number ratio is ≤ the abnormal area ratio threshold, generate a suspected qualified signal. If the number ratio is > the abnormal area ratio threshold, generate an unqualified signal.

3. The method for detecting surface defects of fabrics based on machine vision according to claim 2, characterized in that: The process of obtaining the pilling trend value is as follows: Analyze the pilling area of ​​the abnormal area and conduct pilling test on the normal area to obtain the pilling performance value of the abnormal area and the increase coefficient of the normal area; The pilling trend value is obtained by performing ratio processing on the pilling characteristic value of the abnormal area and the increase and change coefficient of the normal area.

4. The method for detecting surface defects of fabrics based on machine vision according to claim 3, characterized in that: The process of obtaining the pilling performance value of the abnormal area is as follows: Extract the pilling area of ​​the abnormal area, and if it is greater than or equal to the standard value of the pilling area, mark the corresponding pilling as a prominent pilling; Obtaining the average pilling area of ​​all prominent pillings, and calculating the deviation ratio from the standard value of the pilling area to obtain the prominent pilling area deviation ratio; The ratio of the number of prominent pilling and the deviation ratio of the prominent pilling area are summed to obtain the pilling performance value of the abnormal area.

5. The method for detecting surface defects of fabrics based on machine vision according to claim 3, characterized in that: The process of obtaining the mutation coefficient of the normal area is: The pilling test is carried out on the normal area to obtain the pilling grade parameters, which are then compared with the standard pilling grade parameters to obtain the increase coefficient of the normal area.

6. The method for detecting cloth surface defects based on machine vision according to claim 1, characterized in that: The process of outputting regular data of traceability samples includes: Obtain the production time of the unqualified traceability sample and the production time of the strong-impact traceability sample, and divide them into the production time interval of the unqualified traceability sample and the production time interval of the strong-impact traceability sample; Subtract the production time intervals of adjacent unqualified traceability samples to obtain the interval intervals of unqualified traceability samples, calculate the variance of the intervals of all unqualified traceability samples, and obtain the time variance value of strongly influential traceability samples; If it is greater than the time variance threshold of the unqualified traceability sample, an unqualified non-periodic signal is generated; If it is less than or equal to the time variance threshold of the unqualified traceability sample, an unqualified periodic signal is generated.

7. The method for detecting surface defects of fabrics based on machine vision according to claim 6, characterized in that: The process of outputting regular data of traceability samples also includes: Subtract the production time intervals of adjacent strong-impact traceability samples to obtain the interval intervals of the strong-impact traceability samples, calculate the variance of the intervals of all strong-impact traceability samples, and obtain the time variance value of the strong-impact traceability samples; If it is greater than the time variance threshold of the strong impact tracing sample, a strong impact aperiodic signal is generated; If it is less than or equal to the time variance threshold of the strong impact tracing sample, a strong impact periodic signal is generated.

8. A method for detecting surface defects of cloth based on machine vision according to claim 7, characterized in that: The process of obtaining the time of adjacent detections is: Get the mean of the intervals of all unqualified traceability samples, which is the time of adjacent tests for unqualified conditions during the production of the fabric; The mean of the intervals of all strong-impact traceability samples is obtained, which is the time of adjacent inspections of the fabric for strong-impact situations during production.

9. The method for detecting surface defects of fabrics based on machine vision according to claim 8, characterized in that: The process of obtaining the detection time of fabric production equipment is as follows: If the intervals of all strongly-influenced traceability samples are located before the intervals of unqualified traceability samples in terms of time, and the intervals of both are smaller than the preset intervals, the start time of adjacent tests of the strongly-influenced situation shall be recorded as the test time of the fabric production equipment.

10. A cloth surface defect detection system based on machine vision, characterized in that: The system is used to execute the method described in any one of claims 1 to 9, and the system comprises: Preliminary detection module: Analyze the pilling characteristics of the surface image of the fabric, obtain the number ratio of abnormal areas, and determine whether the fabric generates an unqualified signal or a suspected qualified signal; In-depth detection module: Based on the suspected qualified signal, analyze the abnormal marking results of the fabric surface area, obtain the pilling trend value, determine the impact of fabric surface pilling on fabric surface defects, and generate a strong impact signal; Periodicity judgment module: extract all fabric samples with unqualified signals and strong impact signals within the test time, record them as unqualified traceability samples and strong impact traceability samples respectively, obtain the production time, analyze them, and output the regular data of the traceability samples; Optimal detection time determination module: Analyzes the time of adjacent detections for unqualified conditions and strong impact conditions during fabric production to obtain the detection time of fabric production equipment.

Images