图像融合的量化评价方法及实验分析

图像融合是图像信息的综合处理,是将同一对象的一个或更多的图像合成在一幅图像中,以便比原来的任何一幅图像更容易地为人所理解。介绍了图像融合的各种评价方法,包括:熵、交叉熵、交互信息量、均方误差、均方根误差、峰值信噪比。本文利用sym4小波变换的方法进行融合,将融合结果用各种量化评价方法进行了比较,结果表明交叉熵与交互信息量是行之有效的评价方法。

Objective Evaluation and Experimental Analysis of Multisensor Image Fusion

Image fusion is a method of compensative disposal of information, which can fuse one or more images of one object in one picture, so that the image can be understood easily. The methods of the evaluation of image fusion, which includes: entropy, cross entropy, mutual information, mean square error, root mean square error, peak-to-peak signal-to-noise ratio were presented. The method of wavelet sym4 was used to fuse image. By comparing these image fusion methods, the result shows that cross entropy and mutual information are the best of methods of image fusion.