改进稀疏去噪算法下图像自适应融合研究  

Research on Adaptive Image Fusion Based on Improved Sparse Denoising Algorithm

作　　者：蒋澎涛 欧阳建权[2] JIANG Peng-tao;OU YANG Jian-quan(School of Electrical and Information Engineering,Hunan Institute of Traffic Engineering,Hengyang Hunan 421001,China;College of Information Engineering,Xiangtan University,Xiangtan Hunan 411105,China)

机构地区：[1]湖南交通工程学院电气与信息工程学院,湖南衡阳421001 [2]湘潭大学信息工程学院,湖南湘潭411105

出　　处：《计算机仿真》2023年第5期224-227,266,共5页Computer Simulation

基　　金：湖南省教育厅重点科研项目(20A166)。

摘　　要：针对融合后图像易出现信息缺失及细节不清晰等问题,提出基于稀疏去噪的多曝光图像自适应融合算法。引入相异性阈值改进稀疏去噪算法,在图像去噪同时保护图像边缘信息,提升图像去噪效果。融合曲率尺度空间和SIFT算法,构建优化的CSS-SIFT算法用于图像配准,消除拍摄设备自身以及外界环境对图像的不良影响。采用自适应图像块分割构造多曝光图像融合算法,通过自适应图像块分割、图像结构分解、权重图构建以及精细化处理等步骤实现图像融合。实验结果表明,采用所提方法融合后图像具有更好的视觉效果、更高的饱和度、均方差对比度和熵,说明其图像融合效果较好。Due to lack of information and unclear details in images after fusion,this article puts forward an adap-tive fusion algorithm for multi-exposure images based on sparse denoising is proposed.Firstly,the dissimilarity threshold was introduced to improve the sparse denoising algorithm,and protect the edge information while denoising the image,thus improving the image denoising effect.Secondly,the curvature scale space was combined with SIFT algorithm to construct an optimized CSS-SIFT algorithm for image registration,thus eliminating the adverse effects of the shooting equipment and the external environment on the image.Moreover,the multi-exposure image fusion algo-rithm was constructed by adaptive image block segmentation.Finally,the image fusion was completed through adap-tive image block segmentation,structure decomposition,weight map construction and refinement.Experimental results show that the proposed method has better visual effects after image fusion,higher saturation,mean square er-ror contrast as well as entropy.

关 键 词：稀疏去噪 多曝光图像 自适应融合 相异性阈值 

分 类 号：TP393.093[自动化与计算机技术—计算机应用技术]