Image Mosaicing Using Inertial Navigation System for Increasing Match Information in Scene Matching Navigation System

Abstract

Scene matching is one of the key techniques to revise the positioning accumulation error of Inertia Navigation System (INS) for many kinds of unmanned air vehicles (UAVs). When the altitude of aircraft is too low or the ground texture is not rich enough, the traditional scene matching method might fail to get reliable positioning results because of lack of match information. Based on those considerations, we present a new synchronous mosaicing and matching technique based on image mosaicing using INS information to increase match information for scene matching navigation system. In order to achieve image mosaicing, the attitude information measured by INS at imaging instant are used to geometrically transform real-time images to a unified reference coordinate system. Using this scheme, we can eliminate the geometric distortion between real-time images, and then achieve image mosaicing accurately and reliably. In order to ensure the real-time performance of the system, we present a new synchronous mosaicing and matching technique (SMMT). Because of the increase of match information, our method can get more reliable positioning results, and also can reduce the requirement for image texture. Scene matching using many real-time images obtained from the electro-optic pod system clearly justify our method.