Wavelet based progressive mesh compression with random accessibility

Because of self-occlusion, browsing 3D objects is naturally a random access problem. After compression, if only data of visible regions under the current viewing parameter are transmitted and decoded, the network bandwidth and decoding resources can be saved. Almost all state of the art mesh compression schemes do not take random access into account. We propose a wavelet transform based progressive mesh compression scheme which incorporates random accessibility. The mesh is partitioned into many blocks, the detail of each block is encoded independently, then only detail of visible blocks are transmitted. The wavelet coefficients which represent detail are grouped into zerotrees, and each zerotree is encoded independently using a modified SPIHT (set partitioning in hierarchical trees) algorithm. The experimental results indicate the proposed scheme has achieved comparable compression efficiency with PGC, and only about 60% of total bit-stream is required to reconstruct visible regions.