Research On Processing And Editing Of Motion Capture Data

With the rapid development of computer technology and graphics, computeranimation has also developed and penetrated into people’s lives in many fields: games,advertising, medical, education and the military. Motion capture has become the mostimportant means of animation with its unparalleled realism and the powerful editingtechnology of motion capture data. However, due to relatively high cost of motioncapture, simple structure of motion data and can only be used in aparticular environment, the motion capture data must be edited in order to usein the new scenario and make to meet the new needs of users. As a result, the researchon how to process and edit the existing data becomes a hotspot in the motion capturefield.A new motion data conversion method was presented for gaining the entiremotion parameter from original motion data to resolve the problem that originalmotion data is not conducive to the motion editing. Firstly, a simplified humanskeleton model was constructed. The transformation equation of the joint underdifferent coordinates was established using the structural relationship of joints in theskeleton. Then a construction of decomposition method was presented to calculaterotation matrix. At last a human skeleton model was driven by the standard BVHmotion data file that stored the transitional and rotational information of the joints.A motion blending method based on the support phase was presented. Firstly, thecorresponding support phases between two motions were found with support phasedetecting to scale the blending scope. Secondly, the most appropriate transitionsegments between the pair of support phase were found with dynamic time warpingmethod. Thirdly, motion interpolation was used to make a transition between thesegments. The method not only solves the problem of motion synchronization, butalso reduces the amount of computation by narrowing the blending scope. At last, thesmooth and natural and transition was realized efficiently.Based on the above methods, firstly, the experiment of motion capture dataconversion was completed under the MFC environment. Then a motion blendingsystem that can read, blend and play BVH motion capture datas was developed using the OpenGL gr aphics library.