Research On The System Of Generating Character Motion Basing On The Algorithm Of Skeleton Extraction

In the traditional generative process of3D Character Motion, the animators need tospend much time in creating the grid of a character, designing the skeleton of the character,establishing the driven relationship between the grid and the skeleton, and then using the keyframes technology to produce character motion. Developing the system of generatingcharacter motion has become a key area in the research of animation industry, so that we canshorten the manufacture time and simplify the manufacture process. The manufacture processof the generating character motion includes skeleton generation, skeleton movement, skeletonskinning and so on. This paper presents a new skeleton generation method, and study on themotion control technology of character skeleton. For a given3D character mesh, the systemextracts the skeleton automatically, read the standard motion data file, processes the motionretargeting and skinning, at finally generate the motion of the given character.The goal of this research is to design the system of generating character motion, in whichusers only need to give a character mesh and a standard motion data file, then they can get themotion of the character easily. According to the lifelike and real time requirements, weresearch on some closely related problems in the system, and get the following main researchresults.(1) This paper analyzes the method present by Baran and summarizes manyautomatic generation methods of skeleton, and then presents a new methodbasing on grid embedding and grid division. The process of skeleton extraction inthis paper consists of two steps: the embedding of the backbone skeletons, andthe extracting of the branch skeletons. Using Baran’s method to implement theembedding of the backbone skeletons: after the discretization of the given mesh,the algorithm embeds the given skeleton in the best way. This paper presents amethod of extracting of the branch skeletons. It consists of two parts: extractingthe end points of the branch skeletons, and extracting the control joints of thebranch skeletons. By using the improved k-means algorithm we find the endpoints of the branch skeletons. By using grid division idea we find the controljoints of the branch skeletons. Then we integrate all of the joints to form acomplete character skeleton.(2) The paper implements the redirection of the movement to the backbone skeletonsusing the inverse kinematics (IK) method, then moves the branch skeletons usingforward kinematic method. (3) The paper implements the skin animation of the character using the standardlinear blend skinning (LBS) method. At last design a Character MotionGeneration System.