Article ID | Journal | Published Year | Pages | File Type |
---|---|---|---|---|
527501 | Image and Vision Computing | 2007 | 10 Pages |
This paper presents a new adaptive vertex-based coding scheme to improve the coding efficiency of the object shapes of a video sequence. We first smooth out the original contour using the property of perfect 8-connectivity, which reduces the number of vertexes without loss of quality. Then we propose top-down and bottom-up algorithms to select the vertex points. In the top-down scheme, we develop a new adjusting technique to measure the distortion. It is simple for both calculation and implementation, and the reconstructed shape has better quality. In the bottom-up scheme, we develop a new area-based measure that makes the insertion of a new vertex easy. Finally, we propose a scheme that employs multiple dynamic ranges to modify the object-adaptive vertex coding method. It improves the coding performance significantly.