Novel Residual Attitude Estimation Approach Using Georeferenced Satellite Imagery

Abstract

This article presents an efficient novel approach estimating residual attitude based on geometrically corrected (GEO ) satellite images. A technique is presented that uses orbital plane geometry to compute the rotation angle as a function of geographic latitude between GEO image space and radiometrically corrected (RAD ) image space. First, a nonlinear forward model is established that translates the residual errors in roll, pitch, and yaw to scan errors and pixel errors in GEO image space. Subsequently, the inverse problem is solved using Newton's method of nonlinear optimization for estimating residual roll, pitch, and yaw. We demonstrate our results on data products of the high-resolution Indian satellites Cartosat-2E and Cartosat-2F. Further, the superiority of the proposed method is established by comparing it with multiple existing methods in the literature. The R2 measures of goodness of fit for roll, pitch, and yaw estimation based on RAD and GEO products using the proposed method are 0.65, 0.99, and 0.65, respectively; using the existing method, they are 0.074, 0.005, and 0.50.