Research And Implementation Of 3D Facial Expression Capture And Simulation Method Based On Kinect

With the rapid development of virtual technology,users need more and more intelligent and emotional interaction with virtual environment,and facial expressions are also one of the main ways for people to convey their feelings.Now three-dimensional facial expression simulation has become a research and development hotspot in the computer field,because virtual characters with facial expressions can convey more delicate and profound emotions.In recent years,virtual character simulation also emerged in the film industry,video chat system,online games and even advertising production.At the same time,it also makes the animation works with virtual characters have a broader living space in social production and life.This thesis takes this as the main research contents,using Kinect as a collection device,kinect first proposes a multi-feature fusion face expression recognition method based on depth image to recognize the captured face expression,and realizes the dynamic frame recognition of expression to achieve the purpose of improving the recognition rate,and then proposes a three-dimensional face expression simulation method based on the RBF function to solve the problems of distortion and poor fluency during simulation.In view of the noise problem of the depth image obtained by the Kinect sensor,the method of processing the depth image is proposed,and a series of operations such as median filtering and foreground extraction of the collected face are proposed to ensure the effect of improving the face detection rate.Aiming at the problem that the expression recognition rate of users is low,this paper proposes a multi-feature fusion face expression recognition method based on depth image,uses a support vector machine and performs parameter optimization and classification of the fusion features,and compares it with the traditional feature algorithm,the results show that the average recognition rate of expressions is increased from 82.78% to90.02%,and the recognition of dynamic expression frames is completed on this basis.For the traditional method in simulating complex facial expressions,distortion,stuttering and other problems are given a more perfect solution.It mainly includes proposing an improved RBF function facial expression simulation method,optimizing the accuracy of the RBF interpolation method,reducing the occurrence of distortion,and improving the effect of three-dimensional face expression simulation.In order to reduce the problem of jitter during three-dimensional facial expression simulation,this paper also combines the incremental weight method to reduce jitter to enhance stability,and uses Lagrange interpolation method to increase the fluency of expression simulation,making the whole expression simulation process more stable and smooth.The improved RBF expression simulation algorithm in this paper improves the average accuracy of three-dimensional facial expression simulation from 84.76%to 88.13%,which effectively improves the efficiency of expression simulation.On the basis of the above work,this paper designed a three-dimensional face expression recognition and simulation system based on Kinect,which realized the research on real-time recognition and simulation of three-dimensional face targets,and compared the traditional Laplace expression simulation algorithm with the improved RBF expression simulation algorithm proposed in this paper,the average accuracy of the simulation was increased by3.37%,and the average time consumption was reduced by 0.5ms,and the results proved that the improved RBF algorithm proposed in this paper can efficiently complete the three-dimensional face expression simulation.