基于StOMP稀疏方法的高光谱图像目标检测  

Hyperspectral image target detection algorithm based on St OMP sparse representation

作　　者：赵春晖[1] 靖晓昊 李威[1] 

机构地区：[1]哈尔滨工程大学信息与通信工程学院,黑龙江哈尔滨150001

出　　处：《哈尔滨工程大学学报》2015年第7期992-996,共5页Journal of Harbin Engineering University

基　　金：国家自然科学基金资助项目(61405041);黑龙江省自然科学基金重点资助项目(ZD201216);哈尔滨市优秀学科带头人基金资助项目(RC2013XK009003)

摘　　要：稀疏表示方法已经被成功应用于高光谱图像目标检测领域,并且取得了较好的检测效果,但由于高光谱图像往往具有很大的数据量,传统的稀疏检测算法计算成本很高。针对这种情况,提出了应用St OMP算法的高光谱图像稀疏目标检测算法,对求解稀疏系数的步骤进行了改进,减少了此过程中的迭代次数,大幅度降低了运算量,提高了检测速度。使用了2组数据进行仿真实验,结果表明,St OMP算法的应用有效地提高了检测速度与检测精度。This paper proposes a new hyperspectral image（ HSI） target detection method using St OMP reconstruction algorithm. St OMP could be used for the case when the computing cost of traditional sparse detection algorithms is very high because the HSI often has a large amount of data. The sparse representation algorithm has been successfully applied to the field of HSI target detection and it has achieved good results. The new method improves the step of solving sparse coefficients,reducing the number of iterations of this process,which significantly improves the detection efficiency and reduces the computing cost. There are two sets of data in the simulation experiment and the results showed that the St OMP algorithm improves the detection speed and precision effectively.

关 键 词：高光谱图像 目标检测 稀疏表示 St OMP算法 快速运算 

分 类 号：TP751.1[自动化与计算机技术—检测技术与自动化装置]