景象匹配中的几何透视校正和归一化互相关模板匹配

几何失真是使景象匹配性能下降的一个很重要的因素,基于参考图与失真实时图中四对对应点,本文采用透视变换方法,对失真实时图进行校正。对校正后的实时图,本文采用归一化互相关方法,进行模板匹配。仿真实验表明,这两种方法简单易行,校正与匹配结果精确。

Geometric Perspective Rectification and Template Matching by Normalized Cross-Correlation on Image Registration

Geometric distorting degrades the image registration probability efficiently. In this paper, a method of geometric perspective rectification is presented that is based four pairs of corresponding control points in real time and reference map, and transform the entire distorted real time map used this method. An image registration technique is present that is based template matching by normalized cross-correlation. The two algorithms are easy to use and experimental results show them are successful and practical.