哈尔滨工程大学学报2017，Vol.38Issue(4)：640-645,6.DOI:10.11990/jheu.201604006
子空间稀疏表示高光谱异常检测新算法
An anomaly detection algorithm for hyperspectral images using subspace sparse representation
摘要
Abstract
To overcome the low precision of hyperspectral imagery anomaly target detection caused by sparse representation,this paper proposes a new algorithm for anomaly target detection using subspace sparse representation.First,the algorithm optimizes fuzzy C-mean clustering using the particle swarm optimization method.Bands with similar features in the original hyperspectral image are placed in the same class,thereby dividing the whole hyperspectral image into a number of band subspaces but not changing its spatial and spectral features.Then,each subspace is detected by anomaly target detection using a spectral and spatial sparsity divergence index joint weighting.The final target detection result is obtained by overlaying the results of each subspace.Experiments were conducted using real AVIRIS data and the simulation results show that the proposed algorithm achieved very promising anomaly detection performance,with high precision and lower false alarm probability.关键词
高光谱图像/异常目标检测/子空间/稀疏表示/粒子群优化/模糊聚类/稀疏差异指数Key words
hyperspectral imagery/anomaly target detection/subspace/sparse representation/particle swarm optimization/fuzzy clustering/sparsity divergence index分类
信息技术与安全科学引用本文复制引用
成宝芝,赵春晖,张丽丽..子空间稀疏表示高光谱异常检测新算法[J].哈尔滨工程大学学报,2017,38(4):640-645,6.基金项目
国家自然科学基金项目(61571145) （61571145）
黑龙江省博士后基金项目(LBH-Z14062) （LBH-Z14062）
大庆市指导性科技计划(ZD-2016-052) （ZD-2016-052）
大庆师范学院博士基金项目(14ZR07). （14ZR07）
