Research Of Moving Target Detection Technology Based On Machine Vision

Visual information, especially dynamic visual information, takes the most part ofenvironment information that is perceived by human beings. Therefore, perceiving thedynamic visual information in the envirorument has become an important researchdirection in computer vision. The detection and tracking of moving objects is an importantissue in applied vision research filed and it has a wide range of application.The research object of the moving target detection and tracking is video sequence, i.e.image sequence. The purpose of moving target detection is to determine whether there isan object. Furthermore, it also need determine the location, size, shape of the object.Based on the review and analysis of current methods of detecting and tracking movingobject,we lay an emphasis on the research of moving objects detection and tracking inimage sequence with static background.This paper mainly studies on the static context of the moving target detection andtracking. However, various types of moving target detection and tracking algorithms havetheir local advantages and application limitations, so this research will contrast dynamictarget detection of several typical methods. On the basis of the algorithm to carry outin-depth analysis and comparison, the advantages and disadvantages of each algorithm aresummarized. Focus on the various difference method, the results show that the framedifference algorithm for moving target detection of united multi-frame dynamicbackground updating method.Target segmentation, which is now widely used of edge detection, is based on grayimage. In order to use the color information of the image, this paper uses a differentialextremum based on color image edge detection technique. In target tracking stage, thealgorithm uses the mean-shift method combined with active contour. First using activecontour model will be the target of the accurate contour fitting out, but also to the targetlocation, and then use the mean-shift on the target area for tracking. The experiments thatthe difference method in general can provide more complete feature data and morecomprehensive performance.