Super-resolution image reconstruction for Omni-Vision based on POCS

Abstract

The catadioptric omnidirectional sensor, called Omni-Vision, involving capture and automatic interpretation of images, depicts full horizontal panorama 360 degrees view of the surroundings. The field of view band can be easily transformed to a panoramic image and a converntional perspective image. However, it has an intrinsical disadvantage that the angular resolution Omni-Vision is lower than that of conventional video camera. In this paper, a super-resolution method for the Omni-Vision is proposed to reconstruct high resolution panoramic transformed images, in which consecutive images obtained by rotating motion of Omni-Vision are fused using the POCS (project on convex sets) algorithm. Experiment results are also provided to demonstrate the efficiency of the proposed method.