摘要
针对传统的立体匹配算法中存在的低纹理区域和遮挡区域匹配精度低、实时性不好等问题,提出了一种基于图割理论的立体匹配算法。把图像分割成色彩单一的不同区域;计算初始视差图,利用可靠点求取各分割区域的平面模板参数,对模板参教相同的相邻区域进行融合;构造全局能量函数,采用图割算法求取全局能量最小的视差最优分配。实验结果表明,该算法对低纹理区域和遮挡区域均有较好的匹配结果,能够满足高精度、高实时性的要求。
Aiming at the problems that traditional stereo matching algorithm has its drawbacks of low resolution and poor real time, this paper presents a fast stereo matching algorithm based on graph cuts theory. First, the image is divided into different seg- ments of a single color. Then calculating the initial disparity, the plane parameters of single segment are computed only using reli- able pixels, and group neighboring segments with the same plane parameters. Finally construct a global energy function, minimiza- tion of the cost function is achieved by graph cuts. The experimental results indicate that the algorithm has a better performance in low texture regions and occluded portions, and can meet both demands for high resolution and real time.
出处
《微型机与应用》
2010年第10期35-38,共4页
Microcomputer & Its Applications
关键词
立体匹配
图割
图像分割
能量函数
视差
stereo matching
graph cut
image segmentation
energy function
disparity
