Radiation Correction And Performance Evaluation Of Airborne Imaging Spectrometer

Hyperspectral data with high spatial and spectral dimensional information can be used for effective quantitative analysis, making the application of hyperspectral remote sensing, specially airborne hyperspectral remote sensing getting more attention. However, hyperspectral imaging process will be affected by atmospheric composition, resulting in changes of spectral response, and radiation error seriously affected the results of hyperspectral remote sensing quantitative research. In order to make effective use of hyperspectral data object recongnition, exploration and other environmental applications, they must be radiometric corrected to remove the effect of solar radiation, atmospheric transmission, and retrieve object true reflectivity. Meanwhile, considering the complex model in atmospheric composition of the absorption, scattering, reflection, refraction, and many other factors, it is difficult to guarantee its accuracy. How to effectively evaluate the radiation correction method and results, promote the optimization and improve the method of adiation correction, improve the hyperspectral data to quantify the precision of remote sensing application, is a key problem.To solve these problems, this paper based on the analysis of airborne hyperspectral imaging principle and on atmospheric radiation transmission principle designs airborne imaging spectrometer radiation correction method based on MODTRAN radiative transfer model. We also propose a method to evaluate the accuracy of radiation correction, design the corresponding software systems, and demonstrate the effectiveness of the method by real hyperspectral data experience. The main work includes:Frist of all, this paper describes the main processes of atmospheric correction:to alternately iterative calculate the water vapor and visibility and use these parameters to correct the image pixel by pixel, and use experiment to verify the effectiveness of the proposed method.Secondly, design a process to verify the accuracy of the atmospheric correction. Including radiometric performance evaluation parameters based on spectral characteristics and dimensionality reduction.Some spectral matching methods required were analyzed, and describe some dimensional reduction methods to exctract spectral feature to keep local neighborhood information and applied to hyperspectral field. During the process of reduce the dimension to extract some useful information. We also found a method about eigenvalue decomposition of positive defined matrix by dividing the matrix. Based on this method, some dimension reduction method can be applied in hyperspectral data. So the dimensionality reduction method can be effectively applied in hyperspectral data.Finally, test the validity of atmospheric correction method proposed in this paper, calculate the smalarity between the results corrected by the mothod proposed in this paper and the results corrected by the FLAASH mothod. And then use some spectral matching methods to classify the corrected data and the data after dimension reduction. Analyse classification differences. Use the consistent rate of classification results of to characterize the radiation correction accuracy. The results show that the difference can be ignored, this paper’s method has high accuracy and stability, and the data corrected by the method proposed in this article can be apply to variety hyperspectral applications.