The Research On Data Processing For Optical Human Motion Capture

Motion capture is an interdisciplinary emerging field. As an important source of the motion capture data, passive-optical motion capture system, with its easier data acquisition,higher sampling rates and lower limited movement,has been widely applied in film and television entertainment, digital protection of endangered cultures, auxiliary areas, HCI and other related fields. Data processing for motion capture, as important research content,attracts extensive attention of scholars recent years, domestic and overseas, and has obtained some valuable research results. But there still remain some problems in motion capture scattered data process and key-frame extraction of processed data.In this paper, we work on data processing methods,including;(1) In view of the low efficiency of traditional matching method,a Manhattan distance based matching method is proposed, which selects nodes by using the feature of Manhattan distance. Experimental results show that the proposed method efficiently improves matching accuracy and shortens the matching time which proves effectiveness of the proposed method.(2)For limitations of traditional method,a Lagrange interpolation based missing data fitting method is proposed.The approximate trajectory curve can be obtained by using the characteristics of the data.Because of the existence and uniqueness of the interpolation polynomial, the missing data can be fitted. The simulation experiments show the feasibility of the proposed method. In addition, to avoid the error accumulation,noise data is removed directly.(3) In view of the high time complexity of the existing frame abatement algorithm, a Least-squares fit based key-frame extraction method is proposed. In this method, key-frame extraction process is divided into two phases: in the first phase,cut frames that can not to be key frame; in the second phase, use the frame abatement algorithm to extract key-frames. Experimental results show the validity and feasibility of the proposed method.Research results can be applied to studies on data processing for optical human motion capture and support related topics.