摘要
针对高光谱遥感图像波段选择问题,提出一种基于?鱼优化算法(remora optimization algorithm,ROA)的高光谱波段选择方法。为了提升传统?鱼优化算法的全局搜索效率和避免陷入局部极值的能力,采用了立方混沌映射和正弦余弦策略分别对种群进行初始化和迭代下一代个体,并针对波段选择问题提出一种基于密度峰值聚类的分组推荐方法,促进优化算法得到性能更优的波段子集。基于改进混沌?鱼优化算法和分组推荐(group recommendation chaos hybrid remora optimization algorithm,GCHROA)的波段选择方法在2个高光谱遥感数据集上进行分类实验,并与其他先进优化算法进行比较。实验结果表明,该方法能够在所选波段数更小的情况下,获得更好的分类效果,且在Pavia Center数据集上以5%的训练样本,达到97.47%的分类精度。实验证明GCHROA所选择的波段子集保留了数据的重要信息,能够大幅提升高光谱遥感图像后续的处理效率。
Aiming at the problem of band selection of hyperspectral remote-sensing image,this paper proposes a novel band selection method based on the remora optimization algorithm(ROA).To boost ROA’s global search efficiency and counter local extremes,the cubic chaotic mapping and sine cosine strategy are used to facilitate population initialization and individual update.Meanwhile,a density-peak-clustering based group recommendation propels the algorithm to attain optimal band subsets.Finally,a band selection method based on group recommendation and chaos hybrid remora optimization algorithm(GCHROA)is used for classification experiments on two hyperspectral remote-sensing data sets and then compared with other advanced optimization algorithms.The results demonstrate that GCHROA can attain superior classification results with fewer selected bands,which has achieved an accuracy rate of 97.47%on the Pavia Center dataset,with 5%of the training samples.Empirically,the band selected by GCHROA retains essential information,significantly boosting the subsequent processing efficiency of hyperspectral remote sensing images.
作者
李政邦
贾鹤鸣
杜一龙
文昌盛
李永超
LI Zhengbang;JIA Heming;DU Yilong;WEN Changsheng;LI Yongchao(School of Computer Science and Mathematics,Fujian University of Technology,Fuzhou 350118,China;School of Information Engineering,Sanming University,Sanming 365000,China;School of Information and Electrical Engineering,Heilongjiang Bayi Agricultural University,Daqing 163319,China)
出处
《应用科技》
2024年第6期68-74,共7页
Applied Science and Technology
基金
福建省自然科学基金面上项目(2021J011128)。
关键词
高光谱图像
波段选择
元启发式
?鱼优化算法
类可分性
分组推荐
立方混沌映射
正余弦策略
hyperspectral image
band selection
metaheuristic
remora optimization algorithm
class divisibility
group recommendation
cubic chaotic mapping
sine cosine strategy
作者简介
李政邦,男,硕士研究生;贾鹤鸣,男,教授,博士;通信作者:贾鹤鸣,E-mail:jiaheminglucky99@126.com.
