3D Human Animation Generation Based On FastGCN And Prior Model

In recent years,with the development of computer graphics,3D animation technology has been widely used in virtual reality,film animation special effects production and other fields.And one of the three-dimensional human animation is the most difficult.The current method of 3D human body real-time animation often requires high accuracy and high cost.Therefore,a low cost real-time 3D human animation generation method is needed.Firstly,this paper proposes an action recognition method based on FastGCN,and then uses this method to complete the recognition of action types,so that the correct prior model can be selected in turn.A 3D human animation generation method based on prior model is completed.First,this paper preprocesses the data that needs to be sent into the FastGCN training,converts the spatial coordinate information into the form of feature vector through the operation of rotation and partition,and then constructs the graph data structure which accords with the spatiotemporal characteristics of the action information.FastGCN can give the recognition results accurately and quickly.To generate 3D human body animation,this paper constructs a 3D human body model library with rough bone skin animation,combines the model obtained by FastGCN training,uses the spatial coordinates of human body action obtained in real time by RGB-D camera to identify the action.According to the recognition results,the most similar model is selected.In this paper,real human action recognition experiments and three-dimensional human animation generation experiments are carried out.The experimental results show that the action recognition method given in this paper can accurately and quickly give the recognition results,which lays a solid foundation for the subsequent selection of prior models.Compared with other methods,the accuracy of this method has been further improved.The method of 3D human animation generation can accurately and high quality real-time rendering of human animation.