Sift-based modeling and coding of background scenes for multiview soccer video

This paper presents a content-adaptive modeling and coding scheme for static multiview background scenes of soccer games. We discuss a content-adaptive modeling approach for static multiview background imagery that is based on piecewise geometric models of the content. We propose an approach that uses the Scale Invariant Feature Transform (SIFT) to extract the parameters of the geometric models. Moreover, a content-adaptive rendering approach is presented for handling occlusion problems in large baseline scenarios. The experimental results show that our content-adaptive modeling and coding scheme outperforms conventional DIBR schemes.