Method and system for detecting fabric defects on line

Abstract

The application discloses a cloth flaw online detection method and system, and the method comprises the following steps: image acquisition of cloth on a pipeline through an array camera; image feature value coding of image data in a position correlation mode by using a multi-thread and parallel processing mode; and establishment of a cloth feature value position knowledge base. The cloth feature value position knowledge base classifies and manages all image feature values, and image data achieves flat storage and management. For different cloth texture patterns, a training set sample required by a deep learning method is generated through an intelligent texture pattern transformation algorithm, and after pretreatment and training, an accurate cloth flaw identification model is established. The application realizes real-time online detection of cloth on a pipeline, accurately identifies and positions cloth flaws, effectively improves the finished product quality of cloth, and is applied to any cloth production line equipment and system.

Description

TECHNICAL FIELD

[0001] The present application belongs to the technical field of cloth defect online detection, and more particularly to a cloth defect online detection method and system. BACKGROUND

[0002] With the continuous progress of artificial intelligence technology, the related application of industrial production is also more and more widely used. In the production process of cloth, the quality problem of finished product has always been a problem that enterprises pay great attention to. Although enterprises improve the production process and technology, the problem of finished product defect is still inevitable. At present, in most cloth production lines, real-time detection is still relied on manual eye detection. This detection method has many shortcomings. First of all, the experience and proficiency of the detection personnel have a great influence on the detection result; secondly, the labor intensity of manual eye detection is very high, which causes great damage to the human eye, and also lacks consistency and reliability of the detection result of finished product; most importantly, special personnel are needed to complete this task, and the cost paid by the enterprise is large. Therefore, it is of great practical significance to detect the cloth defect on the production line in real time. SUMMARY

[0003] The present application aims to provide a cloth defect online detection method and system, the method acquires images of cloth on the production line through an array camera, adopts a multi-thread and parallel processing mode to encode image feature values in position correlation, and establishes a cloth feature value position knowledge base. The cloth feature value position knowledge base classifies and manages all image feature values, and achieves flat storage and management of image data. For different cloth texture patterns, a training set sample required by a deep learning method is generated through an intelligent texture pattern transformation algorithm, an accurate cloth defect recognition model is established after preprocessing and training, and under the condition of setting a feature threshold, a three-layer comparison filtering algorithm is set to analyze cloth texture patterns in real time, which can make correct evaluation and judgment of cloth defects at high speed and effectively, and detect cloth on the production line in real time.

[0004] In order to solve at least one of the above technical problems, according to an aspect of the present application, a cloth defect online detection method is provided, comprising the following steps:

[0005] S1, for different cloth texture patterns, a training set sample required by a deep learning method is generated through an intelligent texture pattern transformation algorithm, an accurate recognition model based on convolutional neural network is established after preprocessing and training;

[0006] S2, array cameras are arranged in sequence on the production line of cloth, the cloth detection system adopts a multi-thread and parallel processing mode to acquire images of cloth on the production line, and the position of the array camera can be adjusted according to actual conditions.

[0007] S3, the number of display cameras is n, the position number of the camera is i, and 0≤i≤n, according to the position number i of the collection camera, the corresponding collected image data is stored regularly, and the image data is associated and encoded to obtain the position image feature value Vi, thereby establishing a fabric feature value position knowledge base;

[0008] S4, the fabric feature value position knowledge base constructs a two-dimensional corresponding relationship according to the position and the pattern object, classifies and manages all image feature values, and achieves flat storage and management of image data;

[0009] S5, under the condition of setting the matching accuracy a, a group of recognition channels is started for the image data collected by each array camera, which first identifies the category and object of the fabric pattern through the recognition model of the convolutional neural network, and then compares with the standard pattern to detect the error degree m of the collected image data;

[0010] S6, by associating the comparison error degree m of the array camera, it is judged whether the fabric pattern has defects;

[0011] S7, since the display cameras are arranged in sequence, the comparison error degree m of the same fabric pattern can be obtained i , and the average comparison error degree

[0012]

[0013] According to the average comparison error degree , the standard deviation

[0014]

[0015] By standard deviation , the fabric pattern is compared with the standard pattern for the first time, and the fabric position that may have defects is found out;

[0016] S8, on the basis of S7, the correlation function N of the comparison error degree m of the fabric pattern is established:

[0017] N=f(m1,m2,…,mn)(3)

[0018] The standard deviation transfer function S of the comparison error degree m of the fabric pattern N :

[0019]

[0020] In SN When t, the texture pattern of the cloth is compared with the position image feature value, the texture pattern defect of the cloth is further confirmed;

[0021] S9, the method of comparing the position image feature value of the texture pattern of the cloth is extracting the corresponding position image feature value Vi from the cloth feature value position knowledge base, establishing the image feature value matrix V=[V1, V2,..., Vn], and comparing the similarity with the standard image feature value Vt of the corresponding texture pattern of the cloth, to further confirm the existence of defects in the texture pattern of the cloth:

[0022]

[0023] Through the above steps, an accurate cloth defect recognition model can be established, and through the convolutional neural network and the comparison algorithm, the cloth defects on the flow production line can be detected in real time.

[0024] According to another aspect of the application, a cloth defect online detection system is provided, which comprises an image data acquisition module, an intelligent data coding module, a feature value position knowledge base module and a defect detection module.

[0025] The image data acquisition module acquires images of the cloth on the flow production line through an array camera, and the position of the array camera can be flexibly set to ensure accurate, high-speed and complete data acquisition.

[0026] The intelligent data coding module associates the acquired image data with the position of the acquisition point, and regulates and encodes according to the feature value extraction scheme, and the encoding process is completed in a multi-thread, parallel processing mode. The intelligent data coding module can generate training set samples required by the deep learning method through an intelligent texture pattern transformation algorithm.

[0027] The feature value position knowledge base module receives the data processed by the intelligent data coding module, classifies and stores the related data, and manages the data to ensure that the system can be accessed and traversed at high speed.

[0028] The defect detection module can analyze the texture pattern of the cloth in real time online under the condition of setting a characteristic threshold, and can make correct evaluation and judgment on the defects of the cloth at high speed and effectively, and can detect the cloth on the flow production line in real time online.

[0029] According to another aspect of the application, a computer readable storage medium is provided, which stores a computer program, and the program is executed by a processor to realize the steps in the cloth defect online detection method of the application.

[0030] According to another aspect of the present invention, a computer device is provided, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the program, the steps of the online cloth defect detection method of the present invention are implemented.

[0031] Compared with the prior art, the present invention has at least the following beneficial effects:

[0032] The cloth defect online detection method of the present invention realizes real-time online detection of cloth on an assembly line, accurately identifies and locates cloth defects, effectively improves the quality of finished cloth products, and is applicable to equipment and systems of any cloth production line. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings of the embodiments will be briefly introduced below. Obviously, the drawings in the following description only relate to some embodiments of the present invention, but are not intended to limit the present invention.

[0034] Figure 1 Schematic diagram of the system architecture of the present invention. DETAILED DESCRIPTION

[0035] To make the purpose, technical solutions and advantages of the embodiments of the present invention more clear, the technical solutions of the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings of 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.

[0036] Unless otherwise defined, technical or scientific terms used herein shall have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs.

[0037] Example 1:

[0038] like Figure 1 As shown, the present invention provides an online cloth defect detection system, which includes: an image data acquisition module, an intelligent data encoding module, a feature value position knowledge base module and a defect detection module.

[0039] The image data acquisition module acquires images of the cloth on the assembly line through an array camera. The position of the array camera can be flexibly set to ensure accurate, high-speed and complete data acquisition.

[0040] The intelligent data encoding module associates the collected image data with the location of the collection point and performs regulated encoding according to the feature value extraction scheme. The encoding process is completed using multi-threaded and parallel processing. The intelligent data encoding module can generate the training set samples required by the deep learning method through the intelligent texture pattern transformation algorithm.

[0041] The feature value position knowledge base module receives the data processed by the intelligent data coding module, classifies and stores and manages the related data, and ensures that the system can be accessed and traversed at high speed.

[0042] The flaw detection module can analyze the cloth texture pattern in real time on line under the condition of setting a feature threshold value, can make correct evaluation and judgment on the flaws of the cloth at high speed and effectively, and can detect the cloth on the flow production line in real time on line.

[0043] The application also provides a cloth flaw on-line detection method, which comprises the following steps:

[0044] S1, for different cloth texture patterns, a training set sample required by a deep learning method is generated through an intelligent texture pattern transformation algorithm, and after preprocessing and training, an identification model based on a convolutional neural network is established;

[0045] S2, array cameras are arranged in front and back order on a cloth production flow line, a cloth detection system adopts a multi-thread and parallel processing mode to collect images of the cloth on the flow line, and the positions of the array cameras can be adjusted according to actual conditions;

[0046] S3, the number of array cameras is n, the position number of the cameras is i, and 0≤i≤n, according to the position number i of the collection camera, corresponding collected image data are stored regularly, and the image data are associated and coded to obtain position image feature values V, so as to establish a cloth feature value position knowledge base;

[0047] S4, the cloth feature value position knowledge base constructs a two-dimensional corresponding relationship according to positions and pattern objects, classifies and manages all image feature values, and achieves flat storage and management of the image data;

[0048] S5, under the condition of setting a matching accuracy alpha, a group of identification channels is started for the image data collected by each array camera, the group of identification channels first identifies the category and object of the cloth texture pattern through the identification model of the convolutional neural network, then compares with a standard texture pattern, and detects the error degree m of the collected image data;

[0049] S6, the comparison error degree m of the array camera is associated and judged to judge whether the cloth texture pattern has flaws;

[0050] S7, since the array cameras are arranged in front and back order, the comparison error degrees m of the same cloth texture pattern can be obtained i , and the average comparison error degree of the same cloth texture pattern is obtained.

[0051]

[0052] According to the average contrast error degree The standard deviation is obtained

[0053]

[0054] By standard deviation The first comparison filtering is performed on the cloth texture pattern and the standard texture pattern to find the position of the cloth that may have defects;

[0055] S8, on the basis of S7, the correlation function N of the cloth texture pattern contrast error degree m is established:

[0056] N = f (m1, m2, …, m n ) (3)

[0057] The standard deviation transmission function S of the cloth texture pattern contrast error degree m N :

[0058]

[0059] When S N > alpha, the position image characteristic value comparison is performed on the cloth texture pattern to further confirm the cloth texture pattern defects;

[0060] S9, the method of performing position image characteristic value comparison on the cloth texture pattern is to extract the corresponding position image characteristic value Vi from the cloth characteristic value position knowledge base, establish the image characteristic value matrix V = [V1, V2, …, Vn], and perform similarity comparison with the standard image characteristic value Vt of the corresponding cloth texture pattern to further confirm the existence of defects in the cloth texture pattern:

[0061]

[0062] After the above steps, an accurate cloth defect recognition model can be established, and through convolutional neural network and comparison algorithm, the cloth defects on the flow production line can be detected in real time.

[0063] Embodiment 2:

[0064] The computer readable storage medium of the embodiment stores a computer program, which is executed by a processor to realize the steps in the cloth defect online detection method of embodiment 1.

[0065] The computer readable storage medium of the embodiment can be an internal storage unit of the terminal, for example, a hard disk or a memory of the terminal; the computer readable storage medium of the embodiment can also be an external storage device of the terminal, for example, a plug-in hard disk, a smart memory card, a secure digital card, a flash memory card and the like equipped on the terminal; further, the computer readable storage medium can include both the internal storage unit and the external storage device of the terminal.

[0066] The computer readable storage medium of the embodiment is used to store a computer program and other programs and data required by the terminal, and can also be used to temporarily store data that has been output or will be output.

[0067] Embodiment 3

[0068] The computer device of the embodiment includes a memory, a processor and a computer program stored in the memory and executable on the processor, and the processor implements the steps in the cloth defect online detection method of embodiment 1 when executing the program.

[0069] In the embodiment, the processor can be a central processing unit, and can also be other general-purpose processors, digital signal processors, application-specific integrated circuits, ready programmable gate arrays or other programmable logic devices, discrete gates or transistor logic devices, discrete hardware components and the like, the general-purpose processor can be a microprocessor or the processor can also be any conventional processor and the like; the memory can include read-only memories and random access memories, and provide instructions and data for the processor, a part of the memory can also include non-volatile random access memories, for example, the memory can also store device type information.

[0070] Those skilled in the art should understand that the disclosed embodiments can be provided as a method, a system or a computer program product. Therefore, the present solution can be in the form of a hardware embodiment, a software embodiment or an embodiment combining software and hardware aspects. Moreover, the present solution can be in the form of a computer program product implemented on one or more computer usable storage media (including but not limited to magnetic disks and optical storage devices and the like) containing computer usable program code.

[0071] The present solution is described with reference to flowcharts and / or block diagrams of the method and computer program product according to the embodiments of the present solution, and it should be understood that each flow and / or block in the flowcharts and / or block diagrams, and the combination of the flows and / or blocks in the flowcharts and / or block diagrams can be realized by computer program instructions; these computer program instructions can be provided to a general-purpose computer, a special-purpose computer, an embedded processor or other programmable data processing devices to produce a machine, so that the instructions executed by the computer or other programmable data processing devices produce a machine for realizing the functions described in the flowcharts and / or block diagrams. Figure 1one or more processes and / or blocks Figure 1 an apparatus for performing the functions specified in the flowchart or multiple flows and / or blocks.

[0072] These computer program instructions can also be stored in a computer-readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instructions apparatus implementing the flowchart Figure 1 one or more processes and / or blocks Figure 1 an apparatus for performing the functions specified in the flowchart or multiple flows and / or blocks.

[0073] These computer program instructions can also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process so that the instructions executed on the computer or other programmable apparatus provide steps for implementing the flowchart Figure 1 one or more processes and / or blocks Figure 1 an apparatus for performing the functions specified in the flowchart or multiple flows and / or blocks.

[0074] A person of ordinary skill in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by computer programs instructing related hardware, and the programs can be stored in a computer-readable storage medium. When the programs are executed, they can include the processes of the above-mentioned embodiment methods. The storage medium can be a magnetic disc, an optical disc, a read-only memory (ROM) or a random access memory (RAM), etc.

[0075] The examples described in the present application are only used to describe the preferred embodiments of the present application, and do not limit the concept and scope of the present application. Without departing from the design idea of the present application, various modifications and improvements of the technical solutions of the present application made by the engineers and technicians in the art should fall within the protection scope of the present application.

Claims

1. An on-line cloth defect detection method, characterized by, Comprise the following steps: S1, for different cloth texture pattern, through intelligent texture pattern transformation algorithm generates the training set sample needed by deep learning method, after pretreatment and training, establish the recognition model based on convolution neural network; S2, in the production line of cloth, set array camera in the order of before and after, cloth detection system adopts multi-thread, parallel processing mode, image acquisition is carried out to cloth on the production line, the setting of array camera position can be adjusted according to actual situation; S3, the number of array camera is n, the position number of camera is i, and 0≤i≤n, according to the position number i of acquisition camera, the corresponding image data collected is stored regularly, and the image data is associated and coded to obtain position image feature value Vi, so as to establish cloth feature value position knowledge base; S4, the cloth feature value position knowledge base constructs two-dimensional corresponding relationship according to position and pattern object, classifies and manages all image feature values, and realizes flat storage and management of image data; S5, under the condition of setting matching accuracy α, a group of recognition channels is started for the image data collected by each array camera, the group of recognition channels first identifies the category and object of cloth texture pattern through the recognition model of convolution neural network, then compares with the standard texture pattern, and detects the error degree m of the collected image data; S6, the comparison error degree m of array camera is associated and judged to determine whether the cloth texture pattern has defects; S7, obtain the multiple comparison error degrees m of the same piece of cloth texture pattern i , obtain the average comparison error degree of the same piece of cloth texture pattern through the standard deviation First comparison filtering is performed on the cloth texture pattern and the standard texture pattern to find the position of the cloth that may have defects; S8, the correlation function N of cloth texture pattern comparison error degree m is established; N = f(m1, m2,..., m n )(3) Standard deviation transfer function S of cloth texture pattern contrast error degree m N : At S N >α, the position image characteristic value of the texture pattern of the cloth is compared, and the flaws of the texture pattern of the cloth are further confirmed. S9, the method of comparing position image feature value of cloth texture pattern is to extract corresponding position image feature value from cloth feature value position knowledge base, establish image feature value matrix, and compare similarity with standard image feature value of corresponding cloth texture pattern to further confirm whether the cloth texture pattern has defects.

2. The method of claim 1, wherein, Step S7 is specifically as follows: the average contrast error degree of the same piece of cloth texture pattern is obtained The formula is as follows: According to the average contrast error degree The standard deviation is found By standard deviation A first comparison filter is applied to the fabric texture pattern and the standard texture pattern to find the locations of the fabric where defects are likely to exist.

3. The method of claim 2, wherein, Step S8 is as follows: the method of comparing position image feature value of cloth texture pattern is to extract corresponding position image feature value Vi from cloth feature value position knowledge base, establish image feature value matrix V=[V1, V2,..., Vn], and compare similarity with standard image feature value Vt of corresponding cloth texture pattern to further confirm whether the cloth texture pattern has defects:

4. A fabric defect on-line inspection system according to the method of claim 1, said system comprising: Image data acquisition module, intelligent data coding module, feature value position knowledge base module and defect detection module; The image data acquisition module acquires image of cloth on the production line through array camera; The intelligent data coding module associates the collected image data with the position of acquisition point, regulates and encodes according to the feature value extraction scheme, and completes the encoding process in a multi-thread, parallel processing mode, the intelligent data coding module generates the training set sample needed by deep learning method through intelligent texture pattern transformation algorithm; The feature value position knowledge base module receives the data processed by the intelligent data coding module, and classifies and manages the related data; The flaw detection module, under the condition of setting a feature threshold, performs real-time online analysis on the cloth texture pattern by setting a three-layer comparison filtering algorithm, and performs real-time online detection on the cloth on a flow production line.

5. A computer readable storage medium having stored thereon a computer program, characterized in that: The program is executed by the processor to implement the steps in the cloth flaw online detection method of any one of claims 1-3.

6. A computer device comprising a memory, a processor, and a computer program stored on the memory and executable on the processor, characterized in that, The processor executes the program to implement the steps in the cloth flaw online detection method of any one of claims 1-3.

Images