Patterned Fabric Defect Detection And Classification

One of the important ways of patterned fabric defect detection as a guarantee of quality in the production of the textile industry,The traditional manual defect detection has the shortcomings of slow detection speed,low detection accuracy,strong subjectivity and so on.Therefore,it is of far-reaching significance to study the automatic detection system of fabric defects of machine vision to promote the textile industry.The major contents in this paper include automatic fabric defect detection,fabric defect classification algorithm,establish of fabric defect detection system based on machine vision.The main work is as follows:(1)According to the extraction of fabric texture period,putting forward an automatic extraction algorithm of superstition of distance matching functions and their first forward differences.Experimental results show that,the automatic extraction of fabric texture period algorithm works effective and accurate,and has the ability of generalization,deformation invariance and noise invariance.(2)According to patterned fabric defect detection,putting forward an algorithm based on automatic extraction of fabric texture period and local binary pattern.The algorithm of automatic extraction of fabric texture period decides the detection size adaptively,improving detection accuracy.Local binary pattern is invariant to gray and rotation,ensuring the robustness.In addition,automatic extraction of fabric texture period and local binary pattern has low computational complexity.Experimental results show that,the patterned fabric defect detection algorithm performance well on star-patterned fabric and box-patterned fabric.(3)According to patterned fabric defect classification,combing geometric features of binary image of patterned fabric defect and texture features of texture image of the same patterned fabric defect.And random forest is chosen as classifier.Due to the consistency of classification result,the average value of the 10 experimental results as final result,is 83%.(4)Under laboratory conditions,Combing with the software tool and the hardware platform,setting up a fabric defect detection system based on machine vision.Experimental results show that,this system can be used to detect the fabric texture types of common and common types of fabric defects.Detection accuracy and reliability can ensure the application of industrial production.