Automatic Path Planning for Augmented Virtual Environment

摘要

Augmented Virtual Environment (AVE) has been proved efficient in the field of image browsing and video surveillance. Previous works rarely concern on transition between images and videos at different locations. We seek to develop an automatic path planning method to find the way of the transition and willing to help to improve the observer's sense of direction. We firstly propose a viewpoint quality metric which evaluates the quality of viewpoints by measuring the image distortion in video projection. Then we construct a graph for the scene whose nodes denote sample viewpoints and edges denote virtual camera movements. A camera path can be generated automatically using this graph. Finally we apply our method in video surveillance and demonstrate that it improves observers' visual experience.