摘要
本文通过分析PPI算法后续处理存在的问题,引入最大距离法(MD)实现基于PPI的端元自动分类,并将获得的未知端元在波谱库中遍历以匹配最佳地类,最终完成基于PPI端元的矿物信息的自动提取。实验采用美国内华达州Cuprite地区的机载AVIRIS和我国东天山地区的星载Hyperion高光谱遥感数据,利用IDL编程实现矿物信息的自动提取,通过对实验结果的比较分析,验证了本文方法的有效性和实用性。
The Maximum Distance （MD） algorithm and traversal of spectrum library were adopted in this study to solve the further processing of PPI, which usually needs manual selection of endmembers. The PPI endmembers were selected with MD and identified automatically by comparing their spectrum with the samples in spectrum library. Finally, an Airborne Visible/Infrared Imaging Spec- trometer （AVIRIS） hyperspectral image of Cuprite, Nevada, USA and a Hyperion image of East Tianshan, Xinjiang, China were used to validate the methods carried out by the IDL programs, including the MNF, PPI, MD, Traverse, SAM algorithm. The results of two experiments were analyzed, which indicated that the methods mentioned herein are valid and useful.
出处
《测绘科学》
CSCD
北大核心
2013年第4期138-141,共4页
Science of Surveying and Mapping
基金
国家863计划重点项目(2008AA121103)
中国地质大调查项目(1212010816033)
关键词
高光谱遥感
像元纯净指数
最大距离法
端元提取
矿物信息提取
hyperspectral remote sensing
pixel purity index
maximum distance
endmember extraction
mineral information ex-traction
