THE EPIPOLAR GEOMETRY AND APPLICATION OF SPHERICAL PANORAMA

Epipolar geometry is a core issue in computer vision and close-range digital photogrammetry, which could reduce the search range to improve both calculation efficiency and reliability of matching. Currently, epipolar constraint has achieved a wide range of applications based on the traditional pinhole projection images and satellite images. However, panoramic epipolar geometry is different from the other images because of its peculiar geometric modeling. The panoramic epipolar constraint can improve the reliability, accuracy and speed of matching. Nevertheless, this constrain has not been fully applied to the 3D information processing of spherical panorama. The paper focuses on the epipolar geometry for spherical panorama and constructs the epipolar geometry based on the coplanar equation. Then, we derive the mathematical equations and their properties, which provide a theoretical basis for the spherical panoramic epipolar line that could be used in computer vision and the close-range digital photogrammetry. The experiment results show that the epipolar line is effectively validated which can reduce the search range and obtain more useful information.