一种基于轮廓自扩展的GVF算法  

An Improved GVF Algorithm based on Self-Expanding Contour

作　　者：王雅萍[1] 郭雷[1] 

机构地区：[1]西北工业大学自动化学院,陕西西安710072

出　　处：《火力与指挥控制》2009年第4期25-29,共5页Fire Control & Command Control

基　　金：国家自然科学基金资助项目(60175001)

摘　　要：针对Snake算法及GVF算法对初始轮廓覆盖区域敏感性的缺陷,提出一种轮廓自扩展方法,从而使初始轮廓能够有效地覆盖目标所在区域,达到对目标的完整搜索;将其应用到目标跟踪中,根据下一帧中目标与当前帧设置的轮廓的交叉,通过自扩展初始轮廓可以很快将目标划为搜索区域之内。实验表明,基于轮廓自扩展方法的GVF方法可以实现对目标轮廓的完整提取;在目标跟踪中可以自动、有效地跟踪目标。A method for self-expanding contour setting is proposed, which control the area of covering the object. Using this method, the initial contour can cover the whole object, so the entire edge contour is accessibility. And with the application on object tracking, the contour of self-expanding can cover the object perfectly according to the information of the crossing of the initial contour in the current frame and the object contour in the next frame. Experiment shows that the method can extracts the entire contour of the image, and achieve the object tracking automatically and efficiently.

关 键 词：SNAKE算法 轮廓自扩展 GVF 目标跟踪 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]