An efficient video panorama constructing algorithm based on color invariant features

Abstract

Construct a panorama is an efficient way to handle the limitation of camera visual angle. The process of constructing a video panorama can be defined as image registration and fusion of video frames to a composite wide-view image. This paper proposes a video panorama constructing algorithm based on color invariant features. The panorama constructing process contains three steps. At the start, we build the camera motion model based on color invariant features in order to apply our algorithm in a scene where there are illumination, highlights and shadows. After that, for the purpose of reducing redundancy and accelerating speed, key frames are selected for the following image fusion process. At last, the input frames are transformed to a same coordinate to fuse a panorama. Experimental results show that the proposed algorithm is effective and accurate. © 2011 Springer-Verlag.