Research The Detection Algorithm Of Broken Warp Detection System Based On Computer Vision

In the textile industry,the elastic cloth produced by computer knit-stitch machine has a good market prospect.Its core production equipment is computer knit-stitch machine.Its stability,performance of high-speed stitch and the degree of automation are the key indication.It has a very important significance to improve the production efficiency of the enterprise.When sewing machine worked,if it couldn’t stop on time while the Spandex which is thin and transparent,it will make waste product.However,traditional manual visual inspection and mechatronic break detection system are to be time-consuming,laborious and prone to errors.In view of the issues discussed above,this paper makes a general design of its detection system,and through the study of different visual algorithms for image processing and analysis,judge whether it was broken.the main studying contents is as follows:(1)this paper elaborates the problems of the existing mechatronic break detection system,analyzes the problem of cracked ends and its adverse effects.Aims at the existing mechatronic break detection system cannot meet the detection requirements of the computer knit-stitch machine.From three parts to develop,as cracked ends algorithm;cracked ends hardware design and software platform of the cracked ends.(2)Develop the hardware of mechatronic break detection system,In consideration of the play cost and pixel requirements.We assign model MER-500-14GC of the camera and the lens as Computer 8mm.Based on windows and C++platform,designed the camera collecting picture interface and the mechatronic break detection system.Integrate with the requirements for starting and stopping control of the knit-stitch machine,the Industrial Control Treasure USB-4761 is selected.The hardware and the circuit design and construction are completed,and the mechatronic break detection system is realized.(3)Study the break detection system algorithm in theory,use image filtering algorithm and image enhancement technology to remove the influence of noise and lighting on the image,and highlighted the display of the organzine in the image;study the straight-line detection algorithm;use the Hof algorithm to detect the straight-line in the image for the machine learning feature extraction.In the process of break detection,the accuracy and recall rate of organzine detection are greatly improved on the basis of the traditional method.Also use the way studying a large number of annotated data based on the logistic regression and support vector machine and make good progress on supporting vector machine method.(4)Design the break detection platform,analyzed the situation whether there is misjudgment in the process of break detection,analyze and design the break detection algorithm,and make appropriate readjustment in combination with the debugging.Based on support vector machine(SVM)break detection algorithm,a variety of feature extraction methods and image classification algorithms are tried.The accuracy and recall rate of the cross-validation dataset are over 80%and 90%,which reaches the specifications of industrial testing.Through the research of the break detection algorithm,the knit-stitch machine would send out the alarm signal when it creaked ends,and make the machine stopped.Also calibrate the exact position of the creaked ends,avoid the waste products.In the actual production of the factory,its simplicity of operator,the reliability of stop after cracked ends runs up to the industrial standard and brings no waste product.