Extracting Mineral Alteration Information From Remote Sensing Images Based On Genetic Algorithm In Qinghai Lalingzaohuo Region

Mineral resources have become an important pillar for the development of countries.The demand of mineral resources is increasing along with the rapid development of the national economy.Remote sensing as a comprehensive detection technology for ground observation has the advantages of wide range of observation,strong timeliness,and low cost.It has been widely used in the field of geological exploration and the difficulty of geological prospecting can be effectively reduced.Extracting information from remote sensing images has certain advantages in geological prospecting especially in areas with poor natural conditions and inconvenient transportation.The wall-rock alteration is often accompanied with the occurrence of ore bodies.It is an effective way to extract mineral alteration information from remote sensing images,which provides a reliable evidence for the geologists.The traditional remote sensing alteration information extraction method is mainly based on the spectral absorption and reflection characteristics of typical minerals in the research region by repeated comparisons to select bands combination for extracting mineral alteration information.However,the empirical model cannot comprehensively consider the spectral characteristics of minerals in the research region.Meanwhile,the complex symbiotic relationship between minerals will be ignored.Therefore,this paper introduces the principal components analysis(PCA)into genetic algorithms(GAs)to select the optimized features combination of mineralized alterations and extract mineralalteration information by a comprehensive analysis of the geological background and the spectral characteristics of minerals in the research region.First,the multispectral remote sensing features set is established by combining regional geological background and mineral spectral characteristics.Second,the objective function and fitness function of mineral alteration are defined based on GAs and PCA.Then,a remote sensing mineral alteration information extraction method is proposed based on GA.Finally,the proposed method is utilized to select the optimized features combination to achieve efficient mineral alteration information extraction with multispectral images.In this paper,the Qinghai Lalingzaohuo region and Landsat 8 OLI are used as the research area and experimental data,respectively.The original bands in OLI images are combined with band operations to form a multispectral remote sensing feature set of the research area.The mineral alteration objective function and fitness function are defined by considering the geological background of research area and the amount and balance of alteration information around known ore spots.The optimization strategy of the GA integrated with the PCA is used to perform the fitness function and genetic operator operation.Then the optimized features combination is selected for extracting iron stained and hydroxy-bearing alteration in the research area.In experimental verification,the results of GA-based remote sensing alteration extraction method are compared with classical PCA with traditional bands combination,and the predicted target areas for prospecting are verified with the known ore spots.The experimental results show that the proposed remote sensing mineral alteration extraction method has higher agreement with known ore spots than the PCA.It illustrates the effectiveness of the GA-based remote sensing alteration extraction method,which provides a new method and ideas for mineral exploration and mineralization analysis.