基于双目视觉的深度图立体匹配算法研究改进

Improve Algorithm Based on the Depth Map Binocular Stereo Matching

双目立体视觉可以直接模仿人眼与人类视觉的立体感知过程,是计算机视觉研究的核心课题之一。文章主要针对图像的立体匹配算法研究,对区域相关算法进行了分析。在传统的NCC基本灰度相关度测度函数的基础上,将动态规划优化理论引入到匹配中,提高了匹配的速度和准确度。实验表明,本文算法能够满足目标跟踪试验要求,具有重要的理论意义和较高的实用价值。

Binocular stereo vision can directly imitate the human eye and human visual perception of three dimensional, computer vision is the core topic of study. This paper mainly for a three-dimensional image matching algorithm research, the regional correlation algorithm was analyzed. In the traditional NCC basic gray correlation function of the measure on the basis of dynamic programming optimization theory will be introduced to the match, improved matching speed and accuracy. Experiments show that the algorithm can meet the test requirements of target tracking and it is an important theoretical and practical value.