重建误差最优化的运动捕获数据关键帧提取

Keyframe Extraction from Motion Capture Data by Optimal Reconstruction Error

为了实现运动捕获数据的实时压缩,提出一种重建误差最优化的关键帧提取方法.定义重建误差作为关键帧有效性的度量标准;利用帧消减算法得到各消减帧的重建误差,并根据误差大小对消减帧进行排序;最后以消减帧排序为基础计算出重建误差曲线,并依据曲线确定最优压缩率来提取出相应数量的关键帧.实验结果表明,该方法能够高效地从运动捕获数据中提取出具有最优重建误差的关键帧,较好地满足了数据实时压缩的需要.

In order to compress motion capture data in real time, a method is proposed to extract the keyframes with an optimal reconstruction error. By the method, firstly the reconstruction error is defined to evaluate the availability of keyframe. Then a frame decimation algorithm is used to compute the reconstruction error for each decimation frame and sorted into a queue according to the error value. a reconstruction error curve is figured out through the queue and an optimal compression rate is derived from the curve. Finally, a set of keyframes are extracted from the original motion data by the compression rate. Experimental results show that this approach can be applied to effectively extract keyframes with optimal reconstruction error and meet the requirement of real-time data compression.