针对三维运动捕获系统的广泛应用,采样频率越来越高,运动数据库日益增大带来的问题,本文提出多路信号异步采样思想对运动数据进行压缩.理论上论证了在人体运动所固有的客观及主观约束条件下采用本文提出异步采样思想比同步采样能够获取更高压缩比,并针对每一路信号提出了基于三次样条的自适应重采样算法.通过大量运动实验分析了不同运动阶段的数据可压缩性,最高压缩比达到13.24,在单个标记点误差为0.5cm时,数据解压重构毫不影响视觉效果.%With the widespread application of 3D motion capture system, the fact that both sampling frequency and storage for motion database are increasing causes various problems. In this paper, the concept of asynchronous resampling on multi-channel signals is proposed for motion data compression. Theoretical proof on objective and subjective constraint condition of human motion is provided, which contributes to asynchronous sampling for reducing redundancy in order to get higher compression ratio than synchronous sampling based methods. By means of cubic spline function an adaptive asynchronous resampling is achieved for 3d motion data compression. Extensive experiments on motion data show that the proposed algorithm accomplishes high compression ratio at different motion stage,among which the highest one reaches 13.24. Moreover, when the error is 0.5 cm per marker the pose can be reconstructed without visual defects.
展开▼
