The western Guangxi -southeastern Yunnan area is covered with thick vegetation, and hence the extraction of geological structure information from RS images meets with difficulty in that the land cover types constitute the major information in RS images. Land cover types must be removed in extracting geological structure information. In this paper, the components of geological information extracted by the K-L method were used for effective elimination of the land cover information. Geological structure information was extracted by using gray level co-occurrence matrix and directional detection. Field inspection reveals that the precision of the results can be evaluated. The results show that the precision of extracting the geological information, especially the precision of extracting small structure information, is fairly high. Nevertheless, the circular structure information cannot be effectively extracted, and the land cover still affects the results to some extent.%桂西-滇东南是植被高覆盖区域,地质遥感信息相对于地表覆盖(植被)信息而言相对较弱.有效消除地表覆盖信息、突出地质遥感信息是遥感地质专题信息提取的瓶颈之一.本文运用K-L变换提取代表地质构造信息的成分波段,有效滤去地表覆盖信息;利用灰度共生矩阵、定向检测和环形滤波检测相结合的方法,并通过阈值设置,提取地质构造的遥感专题信息,从而达到实现计算机自动提取地质构造信息的目的.经过野外实地验证,结果表明:地质构造信息提取精度较高,特别是一些小的地质构造信息也能有效提取;但是环形地质构造信息没有有效提取,地表覆盖类型对地质构造信息的提取仍有较小影响.
展开▼
