摘要
选择青海太子沟矿区及其外围地区作为研究区,引入支持向量机(SVM)新技术方法,依据U SG S标准光谱曲线来构造特征矩阵,采用LOO(L eave-O ne-O u t)算法来选择核函数及其参数,通过选取一个包括矿化区和非矿化区的训练数据集,直接从ETM多光谱遥感数据中提取出与矿化有关的遥感蚀变信息。经过野外检查验证,本次研究成果取得了良好的实际应用效果。
Taizigou mine and its surrounding areas were selected as the research areas and a new supporting vector method（SVM） was introduced to configure matrix of structure character according to USGS standard spectrum and core function and its parameters were chosen by application of LO0 （Leave-One-Out） arithmetic. The remote sensing alteration information associated with mineralization was extracted straight from ETM multi-spectrum remote sensing data by choosing a training data set including that of mineralization area and barren area. The research has received good result through field confirmation.
出处
《矿产与地质》
2006年第6期656-658,共3页
Mineral Resources and Geology
基金
中国地质调查局国土资源大调查项目(200220140001)资助
关键词
多光谱遥感
蚀变信息提取
支持向量机
ETM数据
LOO算法
multi-spectrum remote sensing, alteration information extraction, supporting vector machine （SVM）, ETM data, LOO arithmetic
