MPEG-4 standard is an object based coding scheme. Shape representation is a new feature in MPEG-4, which specifies objects' opaque characteristics. In this paper, we propose a novel shape coding scheme for MPEG-4 visual standard. The proposed coding scheme segments a binary alpha plane (BAP) into a number of binary alpha blocks (BAB) of varying sizes based on the opaque or transparent nature of a video object. The segmentation is done through a quadtree structure. The two-dimensional redundancy within a non-homogeneous BAB are exploited by removing the consecutive identical rows and columns of pixels. At the end, zigzag scan and arithmetic coding are adopted to achieve better compression. The proposed scheme bypasses the overhead in computation of an intermediate contour representation and its associated conversions. Experimental results show that the proposed coding scheme has an overall better compression performance than some other advanced coding techniques.
展开▼
