景象匹配中的快速畸变校正算法  

A Fast Aberration Calibration Algorithm in Image Registration

作　　者：邓志鹏[1] 杨杰[1] 全勇[1] 

机构地区：[1]上海交通大学图像处理和模式识别研究所,上海200030

出　　处：《上海交通大学学报》2003年第z1期75-78,共4页Journal of Shanghai Jiaotong University

摘　　要：景象匹配在精确制导中具有非常重要的地位,快速实现不同环境下图像畸变校正是其中的一个棘手问题.在常用的方法中,基于区域灰度平均的算法使图像模糊,而基于样条插值的算法计算耗时.提出了基于欧几里德距离的畸变校正方法,该算法不仅满足实时性的要求,而且具有保持图像细节的良好特性.利用傅里叶相关技术对姿态校正后的实时图与基准图进行匹配,整个过程大约耗时0.3s.Image registration has very important place in precise control and guide. And aberration calibration of image, which is fast realized in various environments, is an intractable problem in it. In the common methods, the algorithm based on the average of region gray makes image blur, and the algorithm based on spline interpolation wastes time. This paper put forward a method for the aberration calibration based on Euclid distance. This method not only satisfies the requirement of real time, but also has good characteristic in holding details. The registration of the calibrated realtime image and the base image costs 0.3 seconds.

关 键 词：图像处理 畸变校正 傅里叶相关 景象匹配 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]