基于对偶树复小波域HMT模型的遥感图像融合

摘要

针对遥感图像融合领域的实际应用,提出一种基于对偶树复小波变换与隐马尔可夫树模型结合的图像融合新方法.该算法将分别具有高光谱和高空间分辨率优势的两幅图像进行复小波变换,再对分解后不同频率域的系数选择不同的融合规则处理.采用低频系数加权平均；高频系数先建模,再基于区域能量规则处理的方法,最后完成逆变换得到重构图像.将该算法与其他几种图像融合方法进行比较,实验表明,该算法能够取得较为理想的效果.%For the practical application of the field of remote sensing image fusion, a new image fusion method which combines DT-CWT and HMT is proposed. The DT-CWT decomposition of the two images with the advantages of high spectral and high spatial resolution is conducted respectively with algorithm, then different rules of image fusion for the coefficients in different frequency domains are selected to integrate the images. The method of the weighted average for low-frequency coefficients is adopted. The model for high-frequency coefficients is established, and then the inverse transformation is fulfilled and the reconstructed image is achieved according to the processing method based on regional energy rule. A conclusion that the image fusion algorithm can achieve a more satisfactory result is obtained by comparing the proposed algorithm with other traditional image fusion algorithms.