高级检索

基于模糊判别的快速图切割立体匹配

Fast Stereo Matching Using Reduce-Graph Cuts Founded on Fuzzy Logic

  • 摘要: 针对图切割算法存在匹配时间过长的不足,提出了一种快速的图切割立体匹配方法.选用模糊逻辑提取少量的候选匹配点,然后用这些候选匹配点构造缩减的图,再进行图切割立体匹配.采用标准图像对文中方法进行测试,并与图切割立体算法和动态规划算法进行比较的结果表明,该方法较好地保留了图切割算法匹配的准确性,而运行时间仅为图切割算法的1/35.

     

    Abstract: Graph cut algorithm is usually time-consuming in stereo matching.To overcome this limitation,a fast stereo matching method using reduced-graph cuts is proposed.The proposed method only selects some potential values in the disparity range for each pixel by using fuzzy logic,and then builds a reduced-graph.The proposed method is tested with standard stereo images and compared with graph cut and dynamic programming algorithms.Experimental results show that it retains the advantages of graph cut algorithm,and has less matching time,which is only 1/35 of that of the original graph cut algorithm.

     

/

返回文章
返回