摘要
针对图割法的立体匹配算法耗时多的问题,提出了一种基于S S D和图割的快速立体匹配算法。为了缩小视差搜索范围,缩短匹配时间,先采用区域匹配S S D算法得到初始视差,然后再采用左右一致性校验法去除误匹配点,可以提高初始视差图的质量;在构造能量函数时,把初始视差图中的像素视差作为图割的能量函数的限制项,根据这些限制可以减少不必要的节点,从而减少了计算量,缩短匹配时间。通过实验证明了本文算法在保证匹配图像质量的情况下,能提高匹配效率,减少匹配时间。
For the problem that stereo matching methods based on graph cuts are time consuming,this paper presents a fast stereo matching algorithm based on SSD algorithm and graph cuts algorithm.In order to reduce the parallax hunting zone and matching time,t he SSD matching method is used as a similarity decision criterion to determine the initial disparity map.Then,the left-right conformity verification is adopted to remove mistake matching points.This may enhance the quality of the initial parallax map.When the energy function is constructed,the initial parallax map is used as a limit of function to get the optimization of disparity distribution by graph cuts.Therefore,reducing the node of graphs,the graph capacity and execution time are decreased.At last,it is proved by experiments that this algorithm not only achieve a relatively ideal but also save much time in stereo matching.
出处
《自动化与仪器仪表》
2011年第6期131-132,136,共3页
Automation & Instrumentation
