一种快速双目立体匹配方法

A Fast Stereo Matching Method for Binocular Images

在双目立体视觉中立体匹配是关键技术之一。为了提高匹配速度,提出一种新的立体匹配方法。首先根据极线约束条件限定同名点搜索区域,然后在极线约束区域使用活动轮廓分割出物体区域,进一步缩小匹配点搜索范围。在匹配算法方面,使用了邻域差值模板匹配方法。实验证明,提出的方法能够显著地提高匹配速度,并有较好的匹配精度。

Stereo matching is crucial for the distance measurement with binocular stereo system. Since the two cameras are hardly to be strictly parallel, matching based on epipolar constraint can not be applied directly. Analysing the system＇s inodule, a .new roach- ing method was proposed in this paper. Firstly, the correct loaction area was selected based on the epipolar constraint. Secondly, the background was excluded out of the former area by active counter model. Finaly, the corresponding point~ is matched via feature tem- plate which is formed of neighborhood difference. The ＇experimental results show that the proposed algorithm can improve stereo matching speed and it is more effective in the situation of detecting more than one point on the surface of the same obiect.