Jae-Won Chung

A new vertex-based binary shape coder for high coding efficiency

Abstract This paper proposes a new vertex-based binary shape intra coding scheme that generates a set of vertices reducing shape reconstruction error and efficiently encodes the vertices. The conventional vertex-based coding scheme uses the iterated refinement method (IRM) for vertex selection and the object-adaptive vertex encoding (OAVE) for vertex transmission. In the proposed scheme, we introduce a vertex adjustment method to be applied to the vertices selected by the IRM. The vertex adjustment method significantly reduces the reconstruction error of binary shapes. For efficient transmission of vertices, we also propose 1D vertex detection, vertex reordering and initial vertex encoding methods. Furthermore, a conditional differential chain coding is presented for efficient lossless coding of contours. Experimental results show that the proposed vertex-based intra coding scheme reduces coded bits by 17–29% compared with the conventional vertex-based coding scheme, and by up to 27% compared with the intra coding mode of the context-based arithmetic encoding (CAE) scheme of the MPEG-4 Video Verification Model.