基于方向纹理信息的图像融合  

Image Fusion Based on the Anisotropic Texture Information

作　　者：沙宇恒[1] 丛琳[1] 侯彪[1] 焦李成[1] 

机构地区：[1]西安电子科技大学雷达信号处理重点实验室和智能信息处理研究所,西安710071

出　　处：《电子与信息学报》2007年第3期593-597,共5页Journal of Electronics & Information Technology

基　　金：国家自然科学基金(60133010;60372045);国家"973"计划(2001CB309403);国家"863"计划(2002AA135080)资助课题

摘　　要：鉴于Brushlet分解方向纹理信息的特性,提出一种基于方向纹理信息的图像融合新方法,给出一种基于加权局域能量的融合算子。该方法充分利用已知图像中的高维奇异性信息,有效地保持了各项异性的边缘信息和细节方向纹理信息。对普通图像、航拍图像和医学图像分别进行融合操作,并将融合结果与已知图像的熵和交叉熵作为客观评价指标,与相应小波融合方法进行了比较。仿真结果表明,本文方法融合结果与小波域相应方法比较有明显改善。To take full advantage of the anisotropic texture information and multidimensional singular information in an image, a novel method for image fusion based on anisotropic texture information is presented in this paper. And the local energy operator is provided as the fusion strategies. This method can capture and preserve the anisotropic edge information and detail direction texture of collected images effectively, and it is integrated with brushlet transform to the fusion of natural images, aerial photos and medical images. Objective parameters of evaluating the fusion performance are provided via entropy and cross entropy. Comparisons of performance of Brushlet-based and wavelet-based fusion methods are provided. It is shown that the proposed method has improved the fusion result of wavelet-based ones not only in visual effect but also in objective evaluating parameters.

关 键 词：Brushlet 方向信息检测 小波变换 图像融合 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]