基于统计识别模型的遥感蚀变信息提取

Extraction of Remotely Sensing Alteration Information Based on Statistical Identification Model

遥感数据具有离散性、有限性及非线性关系的数学特征。在专题纹理信息提取分析基础上,以异常区灰度值和矿区控矿构造信息为特征值,利用统计学原理,建立统计识别模型。利用统计识别模型对未知区的特征向量进行识别分类,并应用于吉林东部小西南岔多元成矿靶区预测及矿区附近找矿评价工作中,把提取的蚀变信息与矿山工程地质知识相结合,经野外实地检查与部分探矿工程验证,结果与实际相吻合,获得较好的铜、金矿化显示。

Remote sensing data are discrete, finite and nonlinear. On the basis of the analysis and extraction of thematic texture information, using the grey levels of the abnormal areas and the ore-controlling structure information as the characteristic values, the authors have constructed a statistical identification model based on statistical theory for classing the characteristic vectors of unknown area. It has been applied in the Xiaoxinaneha of east Jilin multivariable metallogenie target forecast and the prospecting and assessment near mining area. Combining extracted altered information with mine engineering geological knowledge, it well shows copper-gold mineralization and the results tallies with fact proved by exploring mineral engineering in the field.