Research on the Reconstruction of Observational Geometry for Scanning Multispectral Image

Most of the scanning multispectral data from domestic land observation satellites only provide the imaging geometry of the scene center, missing the geometry of other pixels. This paper proposes a method for the reconstruction of observational geometry and verify the method using images with observational geometry. The result shows that the relative deviations of the reconstructed observation zenith and azimuth are less than 5%, combined with the Ross-Li BRDF model, it can significantly improve the accuracy of directional surface reflectance and of the calibration coefficients.