说明：  出了一种恢复高质量稠密视差图的立体视觉合作算法．该算法采用基于形态学相似性的自适应加权方 法，迭代地进行局部邻域的自适应聚合和抑制放大，实现高效率和高质量稠密视差图计算．将该算法推广到三目摄像 机立体匹配系统中，通过重建摄像机坐标系实现图像校正，并根据连续性假设和唯一性假设，建立视差空间中的支持 关系和三目摄像机之间的抑制关系．实验结果表明，三目立体合作算法能够得到精确的场景视差映射，并可以实现多 基线方向的遮挡检测．该算法特别适用于由多个廉价摄像机组成的立体视觉系统，在几乎不增加软件和硬件资源的 情况下。就可以得到高质量的稠密视差图．
(Out of a dense disparity map to restore quality stereoscopic cooperative algorithms. The algorithm based on adaptive weighted morphological similarity method iteratively local neighborhood adaptive aggregation and inhibition of amplification to achieve high efficiency and high quality dense disparity map calculation. The algorithm is extended to trinocular stereo matching system, by rebuilding the camera coordinate system for image correction, and assumptions based on continuity and uniqueness of the assumptions in establishing a supportive relationship disparity space and trinocular relationship between inhibition. Experimental results show that the algorithm can cooperate trinocular stereo scene accurate parallax map and can achieve more occlusion detection base line direction. The algorithm is particularly suitable for low-cost cameras consisting of a plurality of three-dimensional vision system, almost no increase in the software and the hardware resources. You can get high-quality)