Expression-driven And Real Time Rendering Of Facial Expressions By Laplacian Transform Grid

In recent decades, with the development of film and games industry, face drivers get the attention of the major areas of research institutions, and gradually become a hot field of computer graphics research.In the field of expression drive, there are two commonly used methods, namely the method of direct expression database and grid-driven approach.And into the past few years, the grid directly driven approach slowly being abandoned by people, it is as though this method has not established expression library, no independent modeling for different performers, etc., but there is a fatal flaw, delay is larger, real bad.The main point of this article from multiple aspects of the model and feature improvements, making real-time grid driving method has been improved, reaching the same way with the effect of the expression library.First, real-time data on the performer’s face is acquired, and to get to the data matrix through decomposition and consolidation methods by analyzing the market all kinds of motion capture device, Kinect depth camera developed by Microsoft Corporation selected as laboratory equipment.This part is the base portion, the resulting matrix provides input data for later work.Secondly, the magnitude of facial muscle changes upon generation of human facial expressions to analyze, by studying the biological characteristics of human facial muscles to determine the changing area and the main force of muscle tissue, according to changes in the magnitude of its size marker feature points, and mapped onto a model for a regional breakdown.Third, By streamlining the algorithm model established model points streamline grid bottlenecks direct drive model approach is that real-time is not high, but the main factors affecting the real-time model is established and in full compliance with human physiology characteristic facial expression changes, so w hen there are many non-critical driving point of time overhead was consumed.The aim is to streamline the model to reduce these non-critical points, in order to improve real-time.Last, By varying the position of the feature point coordinates, using Lapla cian transform algorithm to other model grid points are driven to achieve expression of the simulation results. And the effectiveness and feasibility of this method is verified experiments. The method in film, television, games, and mobile terminal has a v ery wide range of applications.