Extraction of Key-frame from Motion Capture Data Based on Pre-selection and Reconstruction Error Optimization

In order to extract keyframes for animation creation,a method aiming at increasing compression rate and decreasing reconstructed error is proposed.The proposed method consists of two phases,that is,the pre-selection phase and the refinement phase.In the first phase,extremal postures in original motion capture data curve were pre-selected as candidate keyframes.In the second phase,firstly the decimated error was introduced as a metric to measure the importance of keyframe and the reconstructed error was introduced as an optimal objective in the keyframe extraction.In order to get the result keyframes,two algorithms aiming to specified compression rate and reconstructed error were proposed to eliminate less important candidate frames to refine the candidate keyframes,where both compression rate and reconstructed error were optimized.The experimental results demonstrate that the original motion capture data can be compressed at a high ratio,and the proposed approaches meet the requirement of real-time data compression.