Using Shape Representation to Design Panorama Video System

Abstract

The panoramic scene generation is a research active area. The importance of panorama generation is increasing the view angle. Using video to generate the panoramic video, a single camera cannot meet the objective to generate the panoramic view in suitable angle. To solve this problem, two or more video sequences are required to produce the panoramic video. This will give a scene wider than that with a single camera. When the number of cameras is increased, the panoramic video view angle will increase. The goal of this research is to generate a new form of video called panoramic video using two or more videos. The panoramic video is generated using two proposed methods. First, the system of panoramic generation depends on the capturing of samples by using normal cameras and the operation will be done through capturing samples for fixed place at the same time. Secondly, the estimation of motion between sample transformation considers on the challenge in the field which depends on the feature-based method. Feature- based method must be strong enough to extract the fixed features in different circumstances like: distortion, light change, noise and blurring by different transformations (scaling, rotation, translation... etc.) and also the use of the shape representation (region - based method) to resolve the problem of overlapping region. The experimental results show that the system can generate panoramic video from two cameras in different places.