Based On Radiative Transfer Model, Atmospheric Correction Of Airborne Hyperspectral Remote Sensing Image Research

Hyperspectal remote sensing is a burgeoning research area in remote sensing field which develops at the end of the last century. With the maturity of hyperspectal remote sensing and its huge advantage in spectral resolution, hyperspectal remote sensing has been extensively used in many fields, such as oceanography, vegetation ecology, agriculture, geology and atmospheric sciences. During the airborne hyperspectral imaging process, the radiant information of ground object that was received from the sensor is distortion because of the atmosphere scatter and absorption. In order to study hyperspectral remote sensing image quantitatively, the real information is needed to be obtained by correcting the atmosphere and removing the effect of atmosphere on radiant signal. The quantitative application of remote sensing data directly depends on the atmospheric correction method.A rapid atmospheric correction method in airborne hyperspectral image based on the atmospheric radiation transferring model was proposed in this article by analyzing and referring to traditional spectroscopy inversion model and classical atmospheric correction method after the research about the theory of atmospheric radiative transmission. First, preprocess the airborne hyperspectral remote sensing data through radiometric calibration. Secondly, the atmospheric parameters, such as water vapor content and aerosol optical depth which are comparatively important but are not easy to obtain during the atmospheric correction, are iterative inversed based on hyperspectal remote sensing data. In process of papameter retrieval, establish fast look up table based on MODTRAN atmospheric radiation transferring model, Aerosol optical depth is rapidly inversed by Dark Object Methods, and water vapor content is quickly extracted by Atmospheric Precorrected Absorption. Then establish the surface reflectivity look-up table based on the main atmospheric parameters which are inversed by MODTRAN precomputation. Atmosphere is corrected in pixel level according to radiative transfer through the atmosphere. At the same time, the process of look-up table and radiation correction by pixel is built based on MODTRAN precomputation. Atmospheric correction algorithm is optimized parallelly based on multi-core processor. This method in this paper that extract atmospheric parameters from hyperspectral remote sensing image directly, hereby dependence of synchronous data is reduced, radiation error is corrected and the reflectivity of objects is inversed. At last, an airborne hyperspectal remote sensing image rapid atmospheric correction system is designed and developed, and the experiment about atmospheric correction based on real airborne hyperspectral data is completed. The experimental results show that, the method developed in this paper has a characteristic of high speed and high accuracy. Hyperspectal remote sensing image corrected by method in this paper has a good result in both the subjective visual and spectral curve of characteristic pixel.