基于动态规划的快速立体匹配算法

Fast Stereo Matching Algorithm Based on Dynamic Programming

为提高立体匹配算法的匹配速度使其满足实时性要求,同时减少视差图中的条纹现象提高匹配准确率,基于动态规划原理提出一种快速立体匹配算法。利用快速自适应权重累积策略累积匹配成本,通过二维有序表结构加快动态规划的计算速度,采用基于方向滤波的视差后处理方法减少视差图中的条纹现象。实验结果表明,该算法在保证视差图准确的基础上能有效提高立体匹配效率,可应用于实时匹配系统。

A fast stereo matching algorithm based on dynamic programming is proposed to improve the efficiency of stereo matching to meet the real-time requirement and reduce the streaking phenomenon in a disparity map to increase matching accuracy. In this algorithm, the fast adaptive weight aggregation is firstly used to aggregate raw matching costs. Secondly ,the stage of dynamic programming for computation of disparities is accelerated by two dimensional order tables. Finally, a disparity postprocessing method based on oriented filters is employed to reduce the streaking phenomenon in a disparity map. Experimental results show that the proposed algorithm can improve the efficiency of stereo matching and the matching accuracy. It can be applied in real time matching system.