Hyperspectral Metamorphic Lithologies Areas Classified Correspondence Analysis

The technology of remote sensing geological prospecting has become one of the main ways of mineral exploration. Application of hyperspectral remote sensing technology with high spectral resolution in the rock and mineral information recognition and quantitative extraction, greatly improves the rock type zoning map and alteration minerals such as thematic map accuracy, promoting the development of closely related with the mineral exploration of mineral identification and mapping, lithologic mapping technology. Rock spectral absorption feature extraction and lithologic classification is the key for rock and mineral information recognition. The study on absorption characteristics of rock spectra can obtain the recognition characteristics of different rock types in the region. The important reference information for classification of rock is provided for the rapid identification of rock information. Study on the lithology classification technology, extraction, develops new methods of modern mapping techniques for rock and mineral information, acquisition, which is important for regional mineral distribution.However, there are several problems in the application of hyperspectral remote sensing on rock and mineral information recognition and classification:(1) The hyperspectral data easy to be disturbed by many factors in the acquisition process, there are some differences between the acquisition of spectral information and the real spectral information.(2) The physical definition of different rock types in the spectral absorption feature parameters represent is not clear.(3) Conventional multivariate statistical methods including principal component analysis and cluster analysis, on the basis of definition of classification is not clear.This paper takes Duoaoshan copper mine in Heilongjiang province and Luanping area in Hebei province as the study areas, using the principle of information extraction and classification of rock and mineral, wavelet denoising technique based on the rock, to extract the spectral absorption feature parameters, obtain the corresponding analysis result of rock spectra, finally get the rock classification data and rock Duobaoshan area in Heilongjiang province rock classification map. The main research results achieved are as following:(1) Study on the removal of the measured hyperspectral data noise. Wavelet mother function bior4.4and rbio3.9standard were selected to denoise the field of rock spectra of two study regions through comparison of different mother wavelet and wavelet decomposition scale removal of rock spectral noise root mean square error (RMSE).(2) Study on the extraction technology of spectral absorption feature parameters of rock. The continuum removal was used to extract rock spectral absorption feature parameters, including the position of absorption, the depth, width, area, absorption, symmetry of absorption. Among them, Muscovite schist slate (weathered surface) is the most obvious absorption features of all Heilongjiang Duobaoshan area rock, which has the maximum depth and absorption area. The most obvious feature is hornblende biotite gneiss in Hebei province. In addition, Because the field environment effects on the spectral data, different rock types may have the same spectral absorption features. The same type of rock is affected by different weathering degree, which leads to different spectral absorption features.(3) Study on the analysis of rock spectrum of correspondence analysis.The first and second principal component axis represents the main factor of cations, anions and water molecules, and the second principal component axis represents a minority of cations. Rock spectra of Duobaoshan area are divided into4categories. From the first to the fourth class, absorption depth of rock gradually become shallow, absorption area gradually becoming smaller, the absorption peak increasing, clay, iron alteration phenomenon gradually obvious; Rock spectra of Luanping area are classified into3. From the first class to class third, rock reflectivity decreases gradually, dark mineral content increasing, clay, iron alteration phenomenon gradually clear.(4) Study on Hyperion image classification of lithology. By analyzing the vegetation index, the second principal component of vegetation index without soil factors was selected as band R, the third principal component of vegetation index without soil information as band G, the third principal components vegetation index added to soil factors as band B, then they are synthesized. A lithology classification figure was produced in Duobaoshan area. Through the study of geological map overlay that lithologic classification figure, the boundaries of geological map is coincided with that of lithologic classification.