Abstract: Based on Markov Random Field（MRF） model for stereo matching，it introduces a similarity measure that integrates gray with gradient to build an MRF global energy function，and uses improved belief propagation to minimize the global energy function in order to get a disparity map，which is then calibrated by MRF causal systems.Performance indicators are gotten through the Middlebury College Stereo Vision Research website.Experimental results show that accuracy of stereo matching is improved by this method.

Key words: stereo matching, Markov Random Field（MRF） model, belief propagation

摘要： 基于立体匹配的马尔科夫随机场（MRF）模型，构建MRF的全局能量函数的数据项和平滑项，提出一种融合灰度和梯度特征的数据项建立，利用改进的置信度传播方法得到视差图，通过MRF因果系统对其校正。通过Middlebury大学的立体视觉测试平台获得性能指标。实验表明，该方法提高了立体匹配的正确率。

关键词: 立体匹配, 马尔科夫随机场模型, 置信度传播