图像配准的统计数值分析

Statistic Analysis for Image Registration

从图像的灰度信息出发,对联合概率分布进行计算分析,通过最大化联合概率分布在二维平面上的聚集度来完成图像的配准过程,并就图像在完全配准、不完全配准、相互独立的情况下其灰度值的联合概率分布进行了数值统计分析.文中以长波红外图像、中波红外图像、短波红外图像为例进行了仿真.分别对5种配准情况的不同配准测度进行了数值和图表分析对比,分析了不同配准测度之间的评价一致性问题.

One of the image registration methods is to achieve the registration by the maximization of the concentration of joint probability distribution at the 2D plane through computing joint probability distribution with intensity information. This method has been discussed in this paper by statistically analyzing the variety of joint probability distribution of images under the image states of independence, incomplete registration and accurate registration respectively. In addition long wave infrared, medium wave infrared and short wave infrared images have been used in emulating as examples. Different image registration measures according to 5 registration cases have been discussed with the analysis of statistics and graphs. The consistency of different registration measures has been evaluated.