基于梯度光流计算的计算机视觉技术  

Computer vision technology of calculation about optical flow based on grads

作　　者：王述[1] 李军[1] 张宝峰[1] 林贻斌[1] 

机构地区：[1]天津理工学院自动化与能源工程学院,天津300191

出　　处：《天津理工学院学报》2004年第3期62-64,共3页Journal of Tianjin Institute of Technology

基　　金：天津市高校科技发展基金资助项目(020529)

摘　　要：近30年来,计算机视觉技术得到了迅速的发展.然而,巨大的计算量仍是处理计算机视觉问题时的一个十分突出的问题,尤其在处理动态实时计算时更是如此.当今,在计算机硬件性能一定的条件下,解决这一问题就只能从算法上入手,降低算法的复杂性.本文从运动学计算的角度出发给出了如何根据光流来恢复运动和形状的一套简单方法.然后对各种方法进行了比较,从而得出在这一领域里解决实际问题时应如何选取计算方法的观点.最后还介绍了计算机视觉技术的应用及发展前景.In recent thirty years, computer vision technology has made great progress. However, immense data are regarding the hardest problem in dealing computer vision problem, especially in dealing with dynamic and real-time calculation. Today, when computer hardware capability is given, solving this problem can only start with algorithm and reduce complexity of algorithm. In this paper, a series of simple method of restoring movement and shape according to the optical flow from kinematics calculation is introduced. Various of methods are compared. A viewpoint of selecting calculation technique to solve fact questions in this field is deduced. The application and the prospect of computer vision technology are reviewed at last.

关 键 词：梯度光流计算 计算机视觉技术 表观运动 运动学 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术] TP242.62[自动化与计算机技术—计算机科学与技术]