Design And Implementation Of Fabric Defect Detection System Based On Machine Vision

Before being processed,cloths,as raw materials,need to be tested and scored about defects and flaws on its surface.The result will determine whether to use them or reject them.Up to now,most of the domestic textile companies still use human's vision to test the cloth.Human's vision is easy to be affected by the inspector's psychological and physical conditions,and by the surroundings.This method is inefficient,and has a high error rate and missing rate.The traditional method also requires an experienced labour knowing all kinds of characteristic of flaws on different kinds of textiles and different standard scoring systems domestic and abroad.It requires an expensive cost to train such worker and can't be reached under many conditions.Aiming at the shortcomings and the status quo of human's vision in the flaw testing in industry,based on computer vision,by statistics based on image edge detection and support vector machine technique based on multi-classification,using structured and unstructured database technique,this article set up a real time flaw testing system containing flaw feature libraries of various cloths and cloth-quality-evaluation standard libraries.The system realizes image capture,image processing,flaw testing,flaw classifying and cloth-quality evaluation,and tested on a mechanical testing machine.This method performed well,and proved its real-time character.The results demonstrate this testing system has a stronger capability than human beings.Compared with existing flaw testing technique,this system performs better in the following ways:(1)a stronger real time system: images captured by IP camera in real time and transferred using RTSP are delivered to software-side to be processed to extract and classify flaw features.Testing and recognition is under concurrent processing using multithreading.The testing result will be finished as soon as the cloth passes the testing machine.(2)a more accurate testing result: introducing various image detection techniques,this method adds every new flaw into training sample sets to improve recognition rate.(3)adaptive and various evaluation system: using various built-in evaluation system to evaluate flaws in real time to satisfy different users.The method combining IP camera and intelligent testing software in this article reduces dependence on hardware of flaw recognition and implementation costs.This method is able to choose the optimum scheme to improve performance and is flexible enough to improve its testing methods and evaluation system associated with the changing of testing standard of industry.I wish to use this system to improve the qualitytesting technique of textile industry.