A088-0004
Automated Radiation and Geometric Correction Processing Method of Polar-Observing CubeSat BNU-1 Images
Abstract:
This study proposes a set of automated radiation and geometric calibration procedure suitable for BNU-1 images. Histogram matching is used to match the multispectral band data of the BNU-1 satellite with the panchromatic band data. This method assumes that the DN value of the central field of view in the multispectral band is true and corrects the DN value of the edge field of view to achieve relative radiation calibration. Geometric correction is implemented by image registration method based on SIFT algorithm to register BNU-1 level 0 data to MODIS level 1 data in the same period. This method can automatically select the optimal reference image for registration. The method combines the selection of matching points of the whole image with the selection of matching points after the partial image enhancement processing, which can make the number of matching points more and the distribution more uniform, so as to achieve better geometric correction effect.
At present, the method of this study has been applied to the correction of some BNU-1 images. 27 images were selected for evaluation, and the average plane geometric accuracy has been increased from 7.9km to 300m. The problem of inconsistent radiation response between the central field of view and the edge field of view has been effectively improved. In the future, ground-based spectral data from China's 36th Antarctic scientific expedition will be used for radiation verification. Moreover, this technology will be used to realize the automatic production of BNU-1 level 1 products, with a view to providing users with better quality polar observation data.