A comprehensive approach for streaming 3D progressive meshes

Fast and efficient streaming of detailed 3D model over lossy network has long been a challenge, although progressive compression techniques were proposed long time ago. One reason is that packet loss occurring in unreliable networks is highly unpredictable, and leads to connectivity inconsistency and distortions. In this paper, we address this problem by proposing a receiver-based loss tolerance scheme based on a prediction technique. Our method works without introducing protection bits and retransmission. We stream mesh refinement data on reliable and unreliable networks separately so as to reduce the transmission delay as well as to obtain a satisfactory decompression result. The tests indicate that the decompression is completed quickly, suggesting that it is a practical solution. Moreover, the proposed prediction technique achieves a good approximation of the original mesh with low distortion.