摘要
立体匹配是立体视觉中最为关键的一步,对立体视觉具有举足轻重的作用。为了能快速地进行立体匹配,该文结合极线约束和同一点相位相同特性,提出了一种新的基于相位的立体匹配算法。该方法首先根据极线约束确定对应点所在极线,再根据四步相移法得到图像的包裹相位,然后根据合成频率的相位展开方法提取相位,最后在该极线上找到相位相同的对应点。实验证明,与传统的极线最小距离方法相比,大大提高了匹配速度。实验结果验证了算法的有效性。
As the most critical step, stereo matching has play a decisive role on the stereo vision. Ac- cording to the epiplastra constraint and same point of the same phase, a new phase based stereo matching algorithm is proposed in this paper, in order to match feature points more efficiently. It is the polar line of the corresponding point is made by epiplastra constraint, and image package phase is get by the four step phase shift method, and phase is extracted by phase unwrapping method of synthesized frequency, then corresponding points of same phase are found this epiplastra constraint. Experiments show that the run- ning time of this algorithm is much faster than the minimum distance method from polar, and verified the effectiveness of the algorithm.
出处
《工业仪表与自动化装置》
2013年第2期101-104,共4页
Industrial Instrumentation & Automation
基金
黑龙江省教育厅项目"基于多视几何的三维数字扫描系统关键技术研究"(12523051)
关键词
立体匹配
极线约束
相位展开
合成频率
四步相移法
stereo matching
epiplastra constraint
phase unwrapping
synthesized frequency
fourstep phase shift method
