基于MRF和颜色空间的立体图像匹配算法

为了提高立体图像匹配的效率和准确度，对基于马尔可夫随机场（Markov Random Fields，MRF）的立体图像匹配算法进行改进，提出一种基于MRF和颜色空间的立体图像匹配算法。该算法利用颜色距离构建匹配代价函数，以充分利用彩色图像中的颜色信息，并采用加速的信度传播算法进行能量最小化。分别利用灰度信息、RGB信息、HSI信息进行立体匹配实验。实验结果表明，改进后的算法能加快收敛，降低误匹配率。

Stereo Matching Based on MRF and Color Space

For advancing the efficiency and accuracy of stereo matching,improved the stereo matching algorithm based on MRF,then proposed a stereo matching algorithm based on MRF and color space.This algorithm utilized color distance to construct the matching cost function for making full use of the color information of color image,and utilized accelerated belief propagation algorithm to minimize the energy.Intensity,RGB and HSI information were used to make stereo matching experiment.Experiment results demonstrate that the algorithm improved can speed up the convergence and reduce the mismatching rate.