一种新的双眼视觉的立体匹配方法及其应用

A New Two-View Stereo Matching Method and Its Application

立体视觉中最大的困难是对应点的寻找,即需要匹配二幅或更多幅立体图像对之间的特征。本文提出一种新的双眼视觉的立体匹配方法,对积木(BLOCK)一类物体的立体图象匹配十分有效。将这种方法应用于三维物体的识别与定位,实验结果表明,与已有方法相比,本匹配方法简单、有效、匹配快速。文章最后给出用这种方法进行物体识别与精确定位的实验结果。

The major difficulty in stereo vision is the correspondence problem that requires matching features between two or more stereo images. In this paper,a new two-view stereo matching method is presented, which is quite effective to a kind of objects called blocks; We have success fully applied this method to 3-D object recognition and location. Experiment results show that compared with existed matching approaches, this method is simple, effective, rapid. At last, experiment result is given.