Perceptually motivated LSPIHT for motion capture data compression

•Real time precise allocation of available bandwidth.•Shorter encoding and decoding time compared to existing methods.•Scalability and the support of level-of-detail suitable for heterogeneous displays.

Motion capture data compression is necessary for real-time interactive transmission and display of animations. In this work, a highly efficient, fast, scalable method for compressing motion capture clips is proposed by introducing a Linear Set Partitioning In Hierarchical Trees (LSPIHT) algorithm. LSPIHT provides near-optimal reconstruction error given the allocated bits. Our new LSPIHT approach is designed to work with individual channels of motion capture data. Instead of a time-consuming optimization process, we further enhance LSPIHT with a bit rate allocation mechanism based on perceptual factors to distribute the available bandwidth among different channels considering relative visual impacts on the rendered motion quality. User studies are performed to validate the feasibility of our approach. Experimental results show that the proposed method is capable of compressing data at a rate between 40:1 and 60:1 with close to perfect reconstruction quality, which in general outperforms existing methods.

Graphical abstractFigure optionsDownload full-size imageDownload high-quality image (208 K)Download as PowerPoint slide