A dense stereo matching by iterated dynamic programming with using Gabor filters
A dense stereo matching by iterated dynamic programming with using Gabor filters
复制标题
使用 Gabor 滤波器通过迭代动态规划实现密集立体匹配
DOI：
--
复制
发表时间：
2008
期刊：
影响因子：
--
通讯作者：
O. Kapshiy
中科院分区：
文献类型：
--
作者：
Y. Lysak;O. Kapshiy
本文提出了一种求解立体匹配问题的方法。它是基于迭代动态规划的能量泛函的最小化。这种任务的解决方案，其中存在遮挡，弱或周期性纹理区域，为优化过程提供了一些模糊性，传播额外的信息，即地面控制点（gcp）在匹配过程中使用定向空间滤波器，使我们能够获得更精确的密集视差图。通过采用迭代动态规划作为优化技术，沿扫描线和跨扫描线进行优化，可以解决典型的扫描线间不一致问题。因此，将GCP引入到优化过程中，可以提高优化过程的稳定性和效率。
In this paper a method for solving stereo matching problem is presented. It is based on minimization of some energy functional by iterated dynamic programming. The solution such task, where presents of occlusions, weakly or periodic textured regions provides some ambiguity for optimization process, propagating extra information i.e. ground control points (GCPs) in to the matching process by using oriented spatial filters allow us to obtain more accurate dense disparity maps. By employing iterated dynamic programming as optimization technique that performs optimization both along and across scan lines allows a typical inter scan-line inconsistency problem to be solved. Consequently, incorporating GCP in to the optimization process improves its stability end efficiency.