基于多小波域变换和分形维数的图像融合算法  

Image Fusion Algorithm Based on Fractal Dimension and Multi-wavelet Transform Domain

作　　者：沈荣[1] 

机构地区：[1]四川文理学院计算机学院,四川达州635000

出　　处：《西南师范大学学报（自然科学版）》2014年第5期88-94,共7页Journal of Southwest China Normal University(Natural Science Edition)

摘　　要：为了充分利用图像的纹理特征,本文将多小波变换方法和分形理论相结合,提出了一种新的基于多小波变换域方向对比度和分形维数的图像融合算法.该图像融合算法首先通过多小波变换进行原始图像分解,然后采用差分和维数法计算分形维数相应的低频分解系数,建立基于分形维数的低频融合规则,高频部分则根据方向对比度的值通过选择法或加权平均法进行融合计算.该算法对IR图像和可见光图像进行融合实验,采用图像熵、标准偏差以及质量度量这些客观指标评估图像融合的质量.实验结果表明,把分形维数与多小波变换方法相结合进行图像融合处理,图像融合质量和效率都明显提高.To make full use of image texture feature ,multi‐wavelet transform method and fractal theory has been combined ,and a new fusion algorithm is proposed based on wavelet transform domain contrast and fractal dimension of image .First ,the original image is decomposed through multi‐wavelet transform . Second ,calculating the low frequency coefficients of fractal dimension by differential box‐counting .Then , establishing fusion rule of low frequency coefficients based on fractal dimension .Weighted average or se‐lection method is used to calculate the high frequency part based on domain contrast .The experiments u‐sing this algorithm to the fusion process of IR and visual images .Objective index entropy ,standard devia‐tion and quality metrics are applied to evaluate fusion quality .Experimental results show that the combina‐tion of fractal dimension and multi‐wavelet transform can improve process quality and efficiency obviously .

关 键 词：图像融合 分形维数 多小波变换 方向对比度 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]