The main purpose of present stereo matching algorithms is to pursuit high matching precision. These algorithms waste too much time in operating, so they are not suitable for real-time application. In this paper, a saving algorithm, which is the conventional theory and methods in the routing optimization of logistics distribution, is used in stereo matching. A novel stereo matching algorithm, which has a high matching rate and a fast speed, is established. Firstly, the pyramid algorithm is used to obtain the control points. Secondly, the initial disparity map is corrected by the construction methods of the logistics network. Thirdly, the paths of global optimization in the disparity image are corrected by using the real-time of the saving algorithm. Finally, the disparity of global optimization can be obtained. The experiments show that the algorithm increases the reality of stereo matching.%由于目前的立体匹配算法大都以追求高匹配精确度为目的,造成了算法执行时间过长,不利于实时场合的应用.因此,提出了将物流领域中的配送路径优化理论和方法,即节约算法,引入立体匹配技术中,构建一种高匹配精度、高执行速度的立体匹配算法,并将该方法引入到双目立体视觉中.首先用金字塔算法在低分辨率图像上求取控制点,之后结合物流网络的构建理论修正初始视差空间图像,进而利用物流配送路径优化中实时性强的特点,在修正后的视差空间图像上快速找到全局最优路径,从而生成具有全局最优特性的视差.实验结果表明,本方法将大大增强立体视觉技术在实时场合下的适用性.
展开▼