基于小波域分类隐马尔可夫树模型的图像融合算法研究

本文介绍了一种基于小波域分类隐马尔可夫树（Classified Hidden Markov Tree Model，CHMT）模型的图像融合方法。这种方法利用了小波域的一种树形结构Markov链提取小波系数尺度之间的相关性，因而更好的反映了图像空域非平稳变化的特性，在图像融合的预处理阶段最大限度的滤除噪声。同时为配合预处理算法采用了梯度选取规则的融合算法，更好的提高融合后图像的质量。实验结果表明将隐马尔可夫树模型引入图像融合算法后，融合图像的峰值信噪比、信息熵、等效视数等性能指标明显改善。

A research on algorithm about image fusion based on classified Hiden Markov Tree model

In this paper, we introduce an image fusion method which relies on a classification based on wavelet- domain Hidden Markov Tree model （CHMT model}. This method took advantage of a wavelet - domain tree - shaped Markov chain extracts correlation between wavelet coefficients＇ scales, therefore it reflects the changes in non - stationary airspace characteristics of an image better, filter the noise as much as possible in the pre- processing stage. At the same time to tie in with the pre - processing algorithm we used the fusion algorithm based on rules of gradient selection, it could better improve the quality of fused images. Test results show that by introducing the CHMT model, the fused image＇s characteristics such as PSNR, entropy value etc.,, impreved significantly.