一种粒子群优化脉冲耦合神经网络的全色锐化算法  

A Multispectral and Panchromatic Image Fusion Algorithm Based on Particle Swarm Optimization and Pulse-Coupled Neural Network

作　　者：赵志威[1] 付昱凯 杨树文 ZHAO Zhiwei;FU Yukai;YANG Shuwen(Gansu Provincial Computing Center,Lanzhou 730000,China;School of Computer Science,China University of Geosciences,Wuhan 430070,China;School of Surveying,Mapping and Geographic Information,Lanzhou Jiaotong University,Lanzhou 730070,China)

机构地区：[1]甘肃省计算中心,兰州730000 [2]中国地质大学计算机学院,武汉430070 [3]兰州交通大学测绘与地理信息学院,兰州730070

出　　处：《航天返回与遥感》2024年第5期51-63,共13页Spacecraft Recovery & Remote Sensing

基　　金：国家自然科学基金项目(42161069);中央引导地方科技发展资金项目(22ZY1QA005)。

摘　　要：为了进一步降低多光谱与全色影像融合后的光谱和空间信息失真,提高融合质量,文章提出一种粒子群优化脉冲耦合神经网络的多光谱与全色影像融合算法。该算法基于主成分分析和非下采样剪切波搭建融合方法的基础融合框架,在低频系数融合过程中使用细节注射的融合方法,降低非必要的信息注射,从而提高光谱保持度。在融合高频系数时,采用参数自适应的简化脉冲耦合神经网络计算融合权重,并基于粒子群优化算法全局搜索能够获取最佳融合质量的对应参数,以提高空间信息的完整性和清晰度。文章通过三组实验验证提出算法的可行性,并与现有的、经典的融合算法进行对比,实验显示:文章提出的融合算法在三组实验中的光谱角映射均在0.1左右,通用图像质量指数在0.9以上。实验结果表明:该算法不仅能够有效提高全色与多光谱影像的融合质量,而且融合效果稳健,在对比实验中具有最佳的融合性能。In order to further reduce the spectral and spatial distortion of the fused images from multispectral and panchromatic images,and improve the fusion quality,this paper proposes a particle swarm optimization pulse coupled neural network algorithm for multispectral and panchromatic image fusion.Based on the fusion framework of Principal Component Analysis and Non-Subsampled Contourlet Transform,the algorithm uses detail injection fusion method in low-frequency coefficient fusion process to reduce unnecessary information injection and improve spectral preservation.When fusing high-frequency coefficients,a simplified pulse coupled neural network with parameter adaptation is used to calculate the fusion weights,and the corresponding parameters that can obtain the best fusion quality are obtained by global search based on particle swarm optimization algorithm to improve the completeness and clarity of spatial information.The feasibility of the proposed algorithm is verified through three sets of experiments,and compared with existing and classic fusion algorithms.The experiments show that the proposed fusion algorithm has SAM around 0.1 and Q above 0.9 in all three sets of experiments.The experimental results show that the proposed algorithm can not only effectively improve the fusion quality of panchromatic and multispectral images,but also robust,and has the best fusion performance in comparison experiments.

关 键 词：全色与多光谱影像 遥感影像融合 脉冲耦合神经网络 粒子群优化算法 

分 类 号：P237[天文地球—摄影测量与遥感] TP751[天文地球—测绘科学与技术]