A Simple Line Drawing Definition and Transfer Model for Facial Animation Generation

Abstract

The Line Drawing Animation is an active research area in Non-Photorealistic Rendering. Many researchs are focused on the skech abastract, like such as portrait drawing of human and animation generation. However most of the model are too complex to calculate or pay attention to the details which are not stable that are not suitable for realtime transfer for continuous sequence of video. This paper proposes a simple line drawing definition and transfer model with Bézier Curves and the core of the AAM fit parameters. The facial line drawings have seven basic emotions include neutral, happiness, anger, disgust, fear, sadness and surprised. Each of the drawing in a specific model is consisted of a same set of cubic Bézier curves. The proposed model is suitabl for shape conbination anmition. In the experiment, the AAM method is used to get the facial features of the face and then find the nearest combination of the emotion to transfer to the line drawing model. The result shows that the method is simple and fast. Only a few of the parameters are needed to transfer that is suitable to record and communication.

Intelligent Facial Expression Generation Method Based on Artificial Intelligence

“…30 feature points are marked on the front photo, and the marking sequence of the points has a fixed order, which can be referred to Figure 3 By utilizing the live landmark positions of experimental images and estimating the rotation angle based on each facial image, the images in the dataset are divided into several subsets. Figure 4 compares the expression generation performance of our method with pure 3DMM and CycleGan models [10][11][12][13]. It can be seen that the method proposed in this paper can achieve better results under different postures.…”