基于边缘特征和局域方差的图像融合新算法  

Novel image fusion algorithm based on edge feature and local variance

作　　者：段先华[1] 夏加星[1] 

机构地区：[1]江苏科技大学计算机科学与工程学院,江苏镇江212003

出　　处：《江苏科技大学学报（自然科学版）》2012年第2期163-168,共6页Journal of Jiangsu University of Science and Technology:Natural Science Edition

基　　金：国家自然科学基金资助项目(60773172);江苏省自然科学基金资助项目(BK2008411)

摘　　要：针对传统小波变换图像融合过程中存在边缘丢失和纹理信息模糊的问题,设计了一种小波变换的图像融合新算法.对小波分解后的低频系数采用基于边缘特征的选择方法,使得融合图像较好地保留原图像中的边缘轮廓信息;高频系数采用基于局域方差取大准则,这样能够有效地将来自不同图像的纹理特征和细节有机融合起来;为了抑制噪声和控制图像的不稳定性,最后再对低、高频系数的选择结果进行邻域窗口的一致性检验与调整,保证了融合图像的稳定性和连续性.实验结果表明,该算法具有比经典金字塔方法和小波变换方法更好的融合性能.In view of the situation that in the fusion process of the traditional wavelet,there are problems of lost edges and blurred image texture,a novel wavelet-based approach for image fusion is presented.The low-frequency coefficient of wavelet decomposition based on edge feature,the choice method of fused image is better to keep their original image edge profile information.The high-frequency coefficients are selected with local variance which is maximum,so that different image characteristics and details can be combined effectively.In order to overcome the presence of noise and to control image instability,all the coefficients are subsequently performed by a window-based consistency verification process.Experimental results demonstrate that the proposed algorithm has better fusion performance than the classical pyramid and wavelet transform.

关 键 词：图像融合 小波变换 边缘特征 局域方差 一致性检验 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]