Automatic stitching method for spherical panoramic video

Abstract

Panoramic video has become a hot research topic in the field of virtual reality in recent years. Existing research suggests that panoramic video cannot be stitched automatically yet; and to this point, stitching accuracy and time consumption are not satisfactory. After analyzing the coordinate transformation of fish-eye images, this paper presents a solution a panoramic video transforming matrix and proposes the steps that are necessary to calculate it. The paper then puts forward a method for searching for automatic matching points based on Harris corner detection, empirical searching areas and HSV testing. Comparative tests were taken. Our research proves that our automatic stitching algorithm for a panoramic video performs better than the alternatives in terms of accuracy and time consumption.