基于编码方法实现立体视觉中图像的点匹配

Image point correspondence using coded targets

实现自动化的图像匹配一直是立体视觉测量中关键的一步。基于编码方法,对2幅图像中的测量点进行匹配。匹配问题实质是求解空间点在2个(或多个)像平面的对应关系。如果对测量点进行编码,对编码进行身份识别后,即可实现匹配。该编码需在复杂的背景中被有效地识别,并且具有旋转、缩放、变形的无关性。实验表明,该方法匹配正确率高,匹配速度快,是一种解决匹配问题的实用方法。

The central task of closerange photogrammetry is the solution to the automated correspondence problem. The results of using coded targets for automated identification and point matching of marked points for 3D model reconstruction task are presented. Experiment shows that this method has a high precise matching rate and a fast speed. It's a effective solution to the point correspondence.