Hyperspectral Remote Sensing Collaborative Extraction And Recognition Rock And Mineral Information In Heicigou Area

The rise and development of hyperspectral remote sensing,based on the acquisition and reconstruction of fine pixel spectra,directly identify the type,composition,inversion of physical and chemical parameters,this makes the remote sensing from the macro to microscopic detection of a qualitative leap,from the stage of remote sensing geologic interpretation to the quantitative analysis of spectral features of ground objects,it is also an important link of remote sensing geology from the surface application to the quantification development step by step.The paper takes remote sensing geological exploration application as the breakthrough point,extraction of rock and mineral information based on hyperspectral remote sensing,the area is located in the eastern part of the Qinghai-Tibet Plateau,using hyperspectral remote sensing technology,Hyperion and HJ-1A hyperspectral remote sensing data synergy method for rock and mineral information extraction,the mineralized alteration information of rock and metal mineral in Hecigou area was quantitatively extracted,fine classification,target recognition,and confirm the feasibility of Hyperion and HJ-1A hyperspectral remote sensing data synergistic prospecting,and its application in geological prospecting of the superiority.The main achievements and innovation points are as follows:(1).A method of extracting alteration mineralization information in Heizigou area of Qinghai-Tibetan Plateau by using hyperspectral data and co-building spectroscopy was proposed.This paper analyzes the influence of the climate and pressure on the spectral characteristics of the hyperspectral data in the Heicigou area of Qinghai-Tibet Plateau,the advantages and disadvantages of various information extraction and spectral matching classification methods are compared,resampling using the HJ-1A dominant spectral range 450nm-950nm and Hyperion 950nm-2500nm wavelength range,the HJ-Hyperion hyperspectral data is combined with the data to construct the information extraction and target recognition,completed the HJ-1A domestic environment and disaster mitigation of small satellites in geological prospecting for the application attempt,it has great social,economic and strategic significance to carry out the new methods and methods of modern geological mapping in the Qinghai-Tibet Plateau area,and speed up the pace of district adjustment,improve the quality of district tune comprehensively(Chapter 5).(2)A weighted window correlation method based on spectral response of rock and mineral is proposed.Based on the detailed analysis of the correlation between the spectral characteristics of rocks and minerals and their physical and chemical properties,and the traditional spectral matching method,the merits and demerits of the single absorption feature,complete waveform feature,spectral knowledge model and so on are considered synthetically.taking into account the spectral curve of the ground objects in the plateau environmental factors under the influence of the actual ground spectral variation,so as to carry out spectral matching.The basic idea is to segment the spectrum in the whole wavelength range,a high weight value is assigned to the spectral range with diagnostic spectral characteristics,sub-first-order spectral range and the range of environmental impact sensitive areas to give low weight values,in the reflectivity range(0-1)between the normal distribution,combined with a priori knowledge of the calculation and analysis then weight factor training,the selected weighting factors are used to fit the spectral characteristics and angle matching.This method is helpful to improve the spectral classification accuracy of ground objects,the alteration mineralization information was determined,it has important reference value for the regional geological survey and geological prospecting in the study area(Chapter 5).(3)A method is proposed to reduce the band noise of hyperspectral data.In the process of data preprocessing,there is a lot of band noise in the visible light band of HSI data,the relative reflectivity and the relative reflectivity are not smooth enough,there is noise caused by abnormal small reflection peaks,Hyperion visible light and near infrared band banding clear,seriously affecting the image quality and classification applications,the method of distance mean square deviation is implemented in ENVI IDL,the basic idea is to first select a noisy stripe band into a three-dimensional array,find the mean of each band and each column,and then find out the average deviation of the data from the average number of mean squared and the mean square root of the average,finally,gain and offset are obtained to obtain the stripped image array and transform the image array.This method can solve the problem that the normal commercial remote sensing software is not effective to deal with the bad line and the noise of hyperspectral remote sensing data.,affect the image quality and classification results.For the extraction and identification of the following information,for the extraction and identification of the following information,provides high-quality remote sensing data reflecting the true spectral characteristics of the object(Chapter 4).