Regarding difficulties of classical AS stereo matching to meet both of accuracy and speed, an improved algorithm is presented in this paper. Firstly, it modifies matching weight computation by connected notation;Secondly, it divides matc⁃hing region, thus, the pixel in the same small window have the same weight. Finally, it optimizes the algorithm by integral image. Experimental results show that, compared with classical adaptive support method, the improving adaptive support method not only has a more accurate and strong robustness matching result but also cuts the running time.%针对经典的自适应权值( Adaptive Support⁃weight简称AS)立体匹配难以兼顾速度和精度的问题,提出一种改进算法。算法首先结合区域特性(连通性)简化权值计算,然后将待匹配窗划分为若干个小区域,并为同区域内的像素分配相同的权值。最后,利用积分图像优化算法。实验结果表明,与经典的AS算法相比,该算法不仅精度更高,鲁棒性更强且运行时间大大缩短。
展开▼