Research Of Strip Deviation Measurement System Based On Binocular Stereo-vision

With the development of the national economy, strip steel has been applied tomany fields, and there is an increasing demand in the performances of strip steel.However, strip running deviation can lead to serious problems during production. Forthese reasons, a new kind of strip deviation measurement system based on binocularstereo vision is introduced in this paper.In order to meet the special requirements of steel strip production line, this paperimproves the existing system based on the principle of binocular stereo vision system.Firstly, builds a binocular stereo vision system synchronous acquisition image pair byusing web camera loading the infrared technology. Secondly, Image preprocessing iscarried out first in the process of detection and processing for moving steel strip.Then the shortage of the traditional Canny edge detection algorithm is analyzed, thesystem automatically select threshold is loaded on the traditional algorithm in order toextract the edge feature. After finding the edge feature point, using the scale invariantfeature transform (SIFT) matching algorithm find match point. And then using avariety of constraint conditions and combined with the seed region growing methodfor the feature points matching. Then, three-dimensional reconstruction model ofbinocular stereo vision system is used to obtain the matching points of and symmetricpoint of world coordinates. Finally, a complete system has been tested in this paper.The results show that the modified binocular stereo vision system in this paper cancompletely satisfy the index of steel strip production. At the same time, it’s comparedwith performance of the same kind of products and highlights the advantages.The weakness of traditional binocular stereo vision technology which cannotmeet the demand of strip steel is analyzed in this article, and proposed thecorresponding improvement program. The main innovations include the followingaspects:1) It can meet low cost, high precision, accurate and quantitative analysis of thestrip position and offset measurement that transplanting the binocular stereo visiontechnology to strip running deviation detection system.2) The shortage of the traditional Canny edge detection algorithm is analyzed in this paper, and the system automatically select double threshold is loaded on thetraditional algorithm in order to extract the edge feature.3) Using a variety of constraint conditions and combined with the seed regiongrowing method for the feature points matching. It can not only reduce thecomputational burden of the match but also improve the success rate of matching.4) A new algorithm of the strip steel center line was presented by using thefeature points and matching points calculated space symmetry. Through strip steelcenter line fitting and measurements, the positioning and offset measurement areobtained. That reduces the error and improves the accuracy. At the same time, it cannot only get horizontal offset by horizontal curve but also get the strip steel Angledeviation through the analysis of the slope curve.