Research On Atmospheric Correction And Radiometric Calibration For Low Altitude UAV Spectral Image

Compared with traditional aerospace remote sensing,UAV image data has many advantages such as low cost,flexible takeoff and landing,low risk and low flight altitude.Now,it has been used as an important data acquisition method except satellite remote sensing and aerial vehicle remote sensing.The data obtained from the multi / hyperspectral sensors mounted by UAV have been widely used in many fields,such as agricultural,forestry,land and resources investigation,disaster response,camouflage survey,etc.However,the radiation signals of the ground objects under different climate and terrain conditions will be affected by the scattering and absorption of atmospheric molecules.It is necessary to calibrate the atmosphere and calibrate the radiation to obtain the reflectance of the ground objects in order to carry out effective quantitative analysis.However,there are two problems in the traditional atmospheric radiation correction models,such as MODTRAN,6S,SCIATRAN and other commonly used calibration methods.First input parameters are complex and the calculation is slow,and a large amount of field measurement data is needed.Second,the low altitude atmosphere is affected by complex terrain and climate conditions.The description of the model for the low altitude atmosphere is single,which can not meet the accuracy requirements of the UAV spectral data atmospheric correction and radiation calibration.It is of great significance to construct a simplified and fast calibration model for the spectrum data of UAV.Therefore,based on the radiation transmission model of low altitude atmosphere and the calibration technology of UAV,this paper analyzes the vertical distribution mode of low altitude atmosphere.Based on the spectral image of UAV collected in different periods,flight altitude and meteorological conditions,the key image factors of low altitude atmosphere are selected by using the MODTRAN low altitude air transmittance simulation,A simple model of atmospheric fast correction based on MODTRAN is constructed.In addition,the model of radiation distortion correction is built for the radiation distortion caused by single image of UAV.The model of radiation calibration is constructed based on the neural network model,which is based on the neural network model.Then,the simple fast atmospheric correction model is used to calibrate the atmosphere to obtain the accurate surface reflectance.