Depth Map Based Hair Modeling

Hair and hairstyles is an important feature of people which largely affect theimpression to others, so the modeling and simulation of hair is an importantcomponent of virtual human simulation and its methods directly affect the virtualhuman simulation results. The hair itself has a special composition. Between the hairand the hair, the hair and scalp and body exists a very complex interaction. Also hairalso has hardness, color, gloss, thickness, length, shape and other physicalcharacteristics, and can be changed under the action of the man. Based on thecharacteristics of the hair, hair modeling becomes the hotspot and difficulties of theresearch.In this paper, we have study and summarized the existing modeling methods ofthe hair.Finally, we proposal a new way to model the hair. The main researchcontents include the following three points:First, through the study of the existing hair modeling method and analyzes theadvantages and disadvantages of various modeling methods and various kinds ofscenes, we presents a method for hair modeling with the a combination of parametricsurface modeling and generalized cylinder modeling.Methods for parametric surfacemodeling is to build the general model of the hair, and as to the specific local detailsof the hair, the modeling method of generalized cylinder is to be used.Second, the external device of the kinect have the advantage which can obtain theRGB color data flow and also can obtain the corresponding depth informationflow.so we make the contour extraction and filtering from the color image and get thehair each view outline information, at the same time, using the related libraryfunctions to the extract the texture mapping. Through the conversion of coordinates ofthe depth information and computation, we can get the3D point cloud information ofthe related area. In order to further improve the computational efficiency, a method isproposed in this paper which is a sphere containing four bone joint to filter the uselessinformation.Finally, take the second step which has get all the hair information as modelinginput, and make the resulting model to render and display.