一种基于曲波变换与引导滤波增强的图像融合方法  

An Image Fusion Method Based on Curvelet Transform and Guided Filter Enhancement

作　　者：张慧[1] 常莉红 马旭[1] 朱乐 ZHANG Hui;CHANG Lihong;MA Xu;ZHU Le(School of Mathematics and Computer Science,Ningxia Normal Univers ity,Guyuan 756099,Ningxia Hui Autonomous Region,China)

机构地区：[1]宁夏师范学院数学与计算机科学学院

出　　处：《吉林大学学报（理学版）》2020年第1期113-119,共7页Journal of Jilin University:Science Edition

基　　金：宁夏高等学校一流学科建设项目（批准号：NXYLXK2017B11）;固原市科技计划项目（批准号：2019GKGY041）;宁夏高校科技创新平台培育项目

摘　　要：针对可见光图像受光线、天气等影响而成像不清导致图像融合效果差的问题,提出一种基于曲波变换的红外与可见光图像增强融合方法.首先,利用引导滤波对可见光图像细节进行增强;然后,利用曲波变换将增强后的可见光图像和红外线图像分别进行分解,对低频部分取系数最大值,高频部分采用正方形邻域窗口中绝对值和所对应的像素最大值为融合图像的值;最后,利用曲波变换的逆变换得到融合后的结果.实验结果表明,该方法在细节处理、保护边缘及源图像信息的保留上都有良好的效果.Aiming at the problem that the image fusion effec t of visible image was poor due to the influence of light and weather,we proposed a fusion method of infrared and visible image enhancement based on curv elet transform.Firstly,guided filter was used to enhance the details of visibl e image.Secondly,the enhanced visible and infrared images were decomposed resp ectively by using curvelet transform,and the maximum coefficients were taken fo r the low-frequency part,the absolute values and the corresponding pixel maxim um values in the square neighborhood window were used for the fusion image valu es for the high-frequency part.Finally,the fusion result was obtained by usin g inverse transform of curvelet transform.The experimental results show that the proposed method has good effect in detail processing,edge protection and source image information.

关 键 词：图像融合 曲波变换 增强 引导滤波 多尺度分解 

分 类 号：TP39[自动化与计算机技术—计算机应用技术]