高光谱影像波段选择算法研究  

Study on band selection algorithms of hyperspectral image data

作　　者：苏红军[1] 杜培军[2] 盛业华[1] 

机构地区：[1]南京师范大学虚拟地理环境教育部重点实验室,南京210046 [2]中国矿业大学地理信息与遥感科学系,江苏徐州221008

出　　处：《计算机应用研究》2008年第4期1093-1096,共4页Application Research of Computers

基　　金：国家自然科学基金资助项目(40401038);地理空间信息工程国家测绘局重点实验室开放基金资助项目;中国矿业大学科学基金资助项目(D200403)

摘　　要：基于高光谱影像数据的特点,分析了高光谱数据的降维方法。着重探讨了波段选择的若干算法:熵及联合熵、最佳指数因子、自动子空间划分、自适应波段选择、波段指数和最优波段指数等算法。分析了各种算法的有效性、局限性和计算复杂度,并针对波段指数的不足,设计了最优波段指数(OBI)波段选择新算法。最后通过具体的试验,验证了各种算法的性能。Based on the characteristic of the HRS image data, the method of dimensionality reduction was discussed. And then some key algorithms such as joint entropy, optimal index factor, auto-subspace partition, adaptive band selection, band index and optimal band index were analyzed. In addition, the effectivity, shortage and complexity between different algorithms were discussed and compared. According to the shortage of band index, the new band sdection method which named optimal band index （OBI） was proposed. At last, the feasibility of proposed methods was proved by experiment results.

关 键 词：高光谱影像 波段选择 特征提取与选择 

分 类 号：TP751[自动化与计算机技术—检测技术与自动化装置]