A Quasi-Rigorous Approach for the Solution of the Relative Orientation of the Linear Array Satellite Images with Similar Viewing Geometry

In this paper a quasi-rigorous approach for the solution of a two-dimensional relative orientation of two images، acquired during the satellite revisit، is proposed. Unlike the conventional relative orientation methods in frame photography، the proposed approach utilizes both x and y parallaxes for the solution of the relative orientation. Since the revisit images have similar viewing geometry، the ill-conditioning problem arising from the height undulations in the x-parallaxes، does not occur. The proposed algorithm is tested on two afterwards Cartosat IRS-P5 revisit images acquired over a highly mountainous terrain. The accuracy figures indicate a sub-pixel precision of 0. 30 and 0. 47 pixels for the residual x and y parallaxes respectively. The proposed algorithm is suitable for the landslide applications for which highly accurate image co-registration is required