Research On Expression Transfer And 3D Face Reconstruction Based On Morphable Model

Human-computer interaction is an important part of the development of intelligent robots,and facial expression information as a more acceptable form of expression is also a major research field in human-computer interaction.Face recognition is a research hotspot of computer vision.Facial deformation transfer is a development direction of the current intelligent dubbing system.This research direction can be applied to intelligent dubbing systems.It can be constructed by observing the character(target character)image or video.In the three-dimensional model,the robot observes the face of the source character(voice partner)and transmits its facial deformation to the target character.In this way,the robot can realize the mouth-aligned dubbing of the target character by the source character.The more human-like robot language translation system can make the robot not only achieve the accuracy of the language,but also achieve the accuracy of the realistic translation style.This topic constructs a model of expression transfer based on three-dimensional reconstruction grid.First of all,it studies the complexity of face reconstruction and the complexity of simplified model for large poses and face masking.The feature point detection algorithm is studied,and the face area detection is performed based on the HaarLike feature.On this basis,feature point detection is analyzed to verify the effectiveness of the improved algorithm.Secondly,based on facial DLIB feature points to complete the reconstruction of the face model,the subject has studied three different reconstruction models,analyzes the characteristics of each model,and selects a simple reconstruction model suitable for the process of expressing oneself.Again,the expression transfer is implemented on the basis of the reconstruction grid,and the expression transfer part mainly adopts the deformation transformation transfer model based on the affine transformation.This algorithm optimizes the selection of the control point so that the entire deformation transformation is realized under the constraint of the minimum control point.Basic expression transfer.Finally,using the ISOMAP algorithm to build the model's texture vector renders the model to achieve more realistic expression transfer.This model is applied to intelligent service robots to make the human-computer interaction more humanized,more interesting,and robots are more humanized.