Mineral Band Information Extraction And Mapping System Based On Hyperspectral Imagery

As a subdivision spectral imaging technique,hyperspectral remote sensing is an efficient means for geological exploration,ground cover analysis and mineral resource investigation.Hyperspectral mineral identification and mapping are the most important step in remote sensing geological mapping from lithological mapping to mineral mapping,in which the extraction of mineral spectral feature parameters and the types and precision of mineral identification are the core.While the hyperspectral image is a collection of image data and spectral data.We need to study fast and efficient parallel processing methods.In this paper,we design the rule base of alteration mineral recognition based on the design of parametric feature extraction method of alteration mineral stratification identification,and establish an inference engine to develop a hyperspectral image feature extraction and mineral recognition system.The main results are as follows:(1)A parametric spectral absorption feature extraction method based on local spectral continuum removal and polynomial fitting is proposed.This method eliminates background effects and enhances absorption characteristics of the spectrum.Polynomial fitting is used to overcome the influence of spectral noise.On this basis,the method of parameter extraction based on the absorption position,absorption depth and valley width of Al-OH bond minerals and Mg-OH bond minerals was designed.The experimental results show that the proposed method has high precision and can overcome the influence of background and noise.(2)Based on the combination of the main absorption bands and the secondary absorption bands,the mineralogical identification rule base including Al-OH bond minerals and Mg-OH bond minerals was constructed.An alteration mineral reasoning recognition scheme based on expert system is designed and the mapping method is given.(3)We develope a software for feature extraction and recognition of alteration minerals in hyperspectral image.The software is based on multi-core computing architecture.The software integrates local spectral continuum removal and polynomial fitting,parametric feature extraction,mineral reasoning recognition etc.After data testing,we found that the classification and precision of the mineral identification of Al-OH bond minerals and Mg-OH bond minerals are greatly enhanced.