Research On Alteration Anomaly Extraction From Remote Sensing Imagery Based On Fractal Theory

Mineral alteration information extraction from remote sensing imagery is important for ore deposit prediction, however, that obtained using current methods normally contains a lot of influencing information as background. How to separate such background information from mineralization-related alteration anomalies is a prerequisite to effectively make use of remote sensing mineral alteration information in mineral prediction.In this paper, an exploring study on fractal theory applications to the extraction of remote sensing alteration information based on altered minerals’ spectral characteristics in Landsat-7 ETM+ imagery is conducted. According to the extracted ferric contamination alteration anomalies of the research area and the corresponding field investigation, it is indicated that the fractal method is feasible in alteration anomaly information extraction. The main achievements are as follows.1) Analyze the spectral characteristics of ferric contamination alteration minerals(limonite, goethite, and hematite, etc.) based on the USGS standard spectral library and ETM+ remote sensing imagery, determine the bands with diagnostic spectral characteristics in Landsat ETM+ remote sensing imagery, and extract ferric contamination alteration anomalies of the research area with a model of “Mask + PCA + Filtering + Threshold--Segmentation”.2) Study two image decomposition methods based on fractal theory and analyze the different application limits of the fractal models. Explore the applications of the S-A and C-A models to remote sensing alteration anomaly extraction and classification. Comparing with several traditional methods, it is proved that the fractal methods of remote sensing alteration anomaly extraction are feasible.3) Verify the ferric contamination alteration anomalies of the research area extracted with the fractal methods in field investigation, and prove that the fractal methods(S-A and C-A) are effective. Moreover, propose a remote sensing alteration anomaly extraction model based on fractal theory— “PCA+SA+CA” model.