中文摘要：

针对基于区域的立体匹配算法支持窗口难以选择的问题, 提出一种新的自适应窗口区域立体匹配算法. 该区域增长的方法可以动态地获得形状和大小均具有自适应特性的支持窗口, 并通过区域视差范围估计、核心窗口视差近似、动态搜索步距调整等方法使算法得到改进. 研究结果表明: 该算法对于视差不连续和遮挡区域都有着良好的适应性;采用该算法获得的视差图像的准确度高达95%, 且计算时间缩短到5 s以内.

英文摘要：

Aimed at the supported windows selected difficultly in stereo matching algorithm based on local areas, a new stereo matching algorithm of adaptive windows via region growing was proposed. This region growing method can offer appropriate shape and size of supported windows adaptively. In addition, this algorithm is improved by estimating the scope of disparity, approximating the disparity of the core window, adjusting the step dynamically. The results show that this algorithm has good adaptability in discontinuous and occlude areas. It can offer robust disparity estimation image using this algorithm. The precision of the algorithm is more than 95 %, and its calculating time is less than 5 s.