A Novel Dynamic Mesh Sequence Compression Framework for Progressive Streaming

Abstract

In this work, a novel three-dimensional (3D) mesh sequence compression framework suitable for progressive streaming is described. The proposed approach first implements a temporal frame-clustering algorithm based on the curvature of pivot vertex trajectory. Then, a decorrelation method is used to remove the redundancy of data in x, y, and z coordinates. Next, to reduce the amount of mesh sequence data, the vertex motion trajectory data in each cluster is compressed using principal component analysis (PCA). Further, the coefficients of x, y and z coordinates obtained from different principal components are considered as mesh signals, which are processed by a spectral graph wavelet transform(SGWT). Finally, the obtained wavelet coefficients are encoded using CSPECK. By transmitting data on different bit-planes from encoder, a 3D mesh sequence is encoded into a multi-resolution sequence. Experimental results show that the proposed method can realize progressive streaming of mesh sequence. Furthermore, the results also show that the proposed approach outperforms state-of-the-art methods in terms of storage space requirement and minimizing the reconstruction error