Inversion Researches Of Surface Features Based-on Hyper-Spectral Databases

Hyperspectral remote sensing technology is a newly interdiscipline which combine computer science and earth science, which has been widely used in quantitative analysis of remote sensing data, accurate agriculture, geophysics recognition of ecological construction etc. Because different objects have different spectral characteristics, and these characteristics are impact by the composition of object and environment. Therefore, starting from the spectroradiometer characteristics data, study and identify different objects is important for remote sensing applications.This paper on the base of the spectral database, explore the relationship between the surface spectral and its biochemical components, and to build models, and carry on regression of mineral component content and vegetation biochemical component.In terms of spectra data analysis and application, this paper does following jobs:Firstly, in terms of spectral features and preprocessing, according to hyperspectral imaging mechanism, this paper analyzed spectral characteristics; then for the problems of multi-band, data redundancy, high correlation between the band and noise of spectral data, this article propose data pre-processing technology, and eliminate noise, smooth data and reduce the dimension of the data.Secondly, in the view of spectral inversion, carry data mining on spectral data, to explore the relationship between the mineral component and its spectrum, and to explore the impact of biochemical characteristics of the vegetation leaves on its spectral characteristics.Thirdly, for the problems of spectral inversion, study the experience and the semi-empirical model and physical model, especially study the multiple linear regression models and the PROSPECT model.Finally, this paper build inversion model. Use linear regression model explore linear relationship between the mineral composition and its spectral data. Analyze input parameters of PROSPECT model. Apply the model to the prototype system of spectral data analysis. Use this model complete inversion of rocks and plants. These works laid a foundation of the applications of Hyperspectral remote sensing.