基于结构相似度的多波段图像自适应融合算法  

Adaptive Fusion Algorithm for Multi-Band Images Based on Structural Similarity

作　　者：张威[1] 马永斌[1] 陈磊 ZHANG Wei;MA Yong-bin;CHEN Lei(College of Information and Electronic Engineering,Shangqiu Institute of Technology,Shangqiu Henan 476000,China;Henan University,School of Software,Kaifeng Henan 475004,China)

机构地区：[1]商丘工学院信息与电子工程学院,河南商丘476000 [2]河南大学软件学院,河南开封475004

出　　处：《计算机仿真》2023年第11期192-196,共5页Computer Simulation

基　　金：河南省科技公关(222102210141)。

摘　　要：多波段图像的融合性能过差会降低图像内信息的读取率,造成图像信息读取不及时或图像信息缺漏等问题,为了保证多波段图像实时信息的读取能力,规避多幅图像内的信息缺漏现象,提出基于结构相似度的多波段图像自适应融合算法。分析多波段图像的成像原理,并根据该原理涉及的频振现象提取多波段图像的区域特征和细节特征。利用互协函数和尺度常量分别获取区域特征和细节特征的结构相似系数,通过分布式函数依赖发现算法获取两组结构相似系数依赖程度较高的多波段图像的结构相似度,并映射至lαβ空间中,实现多波段图像的自适应融合。仿真结果表明,所提方法在高曝光强度、低曝光强度和正常曝光强度下融合效果均较好、互信息较高。Poor fusion performance of multi band images can reduce the reading rate of information within the im-age,resulting in problems such as delayed or missing image information.In order to ensure the real-time reading abil-ity of multi band images and avoid information leakage within multiple images,a multi band image adaptive fusion al-gorithm based on structural similarity is proposed.At first,the imaging principle of multi-band image was analyzed,and then regional and detailed features of multi-band image were extracted through the frequency vibration involved in the principle.Respectively,cross-covariance function and scale constant were used to calculate the structural simi-larity coefficients of regional features and detail features.Moreover,the structural similarity between the multi-band images with high dependence of structural similarity coefficients was obtained by the distributed function dependency discovery algorithm,and mapped to laβspace.Finally,the adaptive fusion for multi-band images was achieved.Sim-ulation results show that the proposed method has better fusion and higher mutual information under high,low,and normalexposureintensities.

关 键 词：多波段图像 动态视觉传感器 特征提取 结构相似度 自适应融合 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]