Low-complexity acquisition of 180-degree panoramic video using early data reduction

Abstract

This paper presents a low-complexity video acquisition system based on multiple cameras and processors for generating panoramic images with spatial resolutions beyond ultra-high definition and field of view (FoV) of 180 degrees. The UHD 180-degree images are formed by stitching nine high-definition images captured from individual cameras, mounted in regular geometric structure, covering a FoV of 180 degrees around the acquisition viewpoint. An experimental study is carried out to reduce the computational complexity required by feature extraction and matching which are the most complex tasks in the image stitching process. Based on an early data reduction approach, it is shown that good quality stitching can be achieved with low computational complexity. Overall the proposed system is capable of capturing 180-degree panoramic images of ultra high definition using low-cost cameras and processors, which makes it suitable for applications with highly demanding resolution requirements.