基于多特征的红外与可见光图像融合  

Infrared and visible light image fusion based on multiple features

作　　者：刘明君 董增寿[2] LIU Mingjun;DONG Zengshou(Electronics Department,Xinzhou Teachers University,Xinzhou Shanxi 034000,China;Taiyuan University of Science and Technology,Taiyuan Shanxi 030024,China)

机构地区：[1]忻州师范学院电子系,山西忻州034000 [2]太原科技大学电子信息工程学院,太原030024

出　　处：《激光杂志》2019年第10期81-85,共5页Laser Journal

基　　金：国家自然科学基金资助(No.41272374);山西省高等学校教学改革创新项目(No.J2018162);忻州师范学院院级课堂教学改革专题研究项目资助(No.JGZT201601);山西省教育科学“十三五”规划项目(No.GH—17058)

摘　　要：采用多尺度变换方法融合红外与可见光图像后,图像融合结果整体场景对比度低,细节与纹理不够清晰,提出基于多特征的红外与可见光图像融合方法。首先提取红外与可见光两种源图像中的边缘特征、相关信号强度比特征及平均梯度特征,通过决策模块依据信号强度比特征和边缘特征确定多分辨率分析系数,然后基于该系数融合模块依据平均梯度特征,采用多分辨率奇异值分解方法融合图像低频子带,利用压缩感知方法融合高频子带背景区域,通过对背景区域高频、低频子带进行CS逆变换获取最终融合图像,最后仿真测试结果表明,该方法的行红外与可见光图像融合的标准差、平均梯度以及边缘保持度等指标都优于对比方法,融合后图像对比度和空间域内活跃度较好、图像细节较明显、图像内容更丰富。After the multi -scale transformation method is used to fuse infrared and visible images,the overall scene contrast of the image fusion results is low,the details and texture are not clear enough,and a multi-feature in- frared and visible image fusion method is proposed. The edge characteristics,the intensity ratio and the average gradi- ent characteristics of the infrared and visible light source images are extracted. The multi-resolution analysis coefficient is determined by the decision module based on the signal strength ratio characteristics and the edge characteristics. Then,based on the average gradient characteristics of the coefficient fusion module,the low frequency subbands are fused by multi-resolution singular value decomposition. The background areas of high frequency subbands are fused by compression perception. And the final fusion image is obtained by CS inverse transformation of high frequency and low frequency subbands in the background area. The results show that the standard deviation,average gradient and edge retention of infrared and visible light images are superior to the contrast method. After fusion,the image contrast and spatial domain activity are better,the image details are more obvious,and the image content is more abundant.

关 键 词：多特征 红外图像 可见光图像 图像融合 低频子带 高频子带 

分 类 号：TN751[电子电信—电路与系统]