Research Of Human Motion Capture Data Distortion Recovery

Motion capture?MOCAP?is an important technique that is widely used in many areas such as computer animation,film industry,and physical training and so on.However,even with professional MOCAP system,the data corrupting problem always occurs when practically gathering data.Therefore,MOCAP data recovery is an essential preprocessing step for MOCAP data before applications and it is still a challenging task due to the complexity and diversity of human motion.In this paper,the issue of human motion capture data recovery has been studied by analyzing and utilizing the characteristic of human motion and the property of motion capture data.The main contents and achievements are as follows:To address the issue of denoising long motion sequence with complex semantics,sparse subspace clustering is utilized to segment the motion data into separated semantic segments due to the high-dimensional characteristics of human motion capture data.In the process of semantic segmentation,in order to make the segmentation result more accurate,the semantic information is emphasized by coordinate transformation of MOCAP data in different processing stages.Meanwhile,the smooth constraint is enforced to the representation coefficient due to the sequential feature of MOCAP data in time.Since the sparse subspace clustering algorithm is robust to noise,the segmentation also removes part of the noise from the MOCAP data and makes the remaining noise sparser.The MOCAP data is then reconstructed with low rank representation to remove the remaining noise in the MOCAP data.Compared with the traditional low-rank decomposition theory,we reconstructed the data by using the data itself with the low rank representation instead of directly using the low-rank constraint of the data to make the reconstructed data preserve the information of spatio-temporal structural.The smooth and symmetric constraints are enforced to the representation coefficients to make the algorithm more robust to noise.Four motion sequences with complex semantics are selected from the database for simulation,compared with the existing algorithms,the results of the experiment show that the proposed algorithm is more robust and has higher recovery accuracy than them.For the situation that the adjacent markers of Motion Capture?MOCAP?data missing for a period of time due to lights and other factors when practically gathering data,a new MOCAP data recovery algorithm is proposed by using the latent correlation and the skeleton constraint in MOCAP data.The algorithm firstly transforms the MOCAP data to represents the changes of the relative position of adjacent markers to acquire the skeleton constraint term.Then the sparse representation and the skeleton constraint term are used for training dictionary which can preserve the structure of motion capture data and the dictionary is then utilized to recovery missing data.The experiment results show that the algorithm can improve the recovery accuracy of the coordinates of the missing markers and increase the bone length recovery accuracy to 10^-4cm,and verify the feasibility and effectiveness of the algorithm.