On Virtual Human Emotion Model And Technology Fulfillment

It is the task for artificial intelligence to give machine the human intelligence such as thinking, reasoning and decision. Endowing machine a life is a hot research field of artificial life, which includes life evolution, multiplication, self-adaptive, self-organization. Artificial emotion gives machine various sensibility like laughing, angry, sorrow and happiness. It is impossible for artificial emotion to be separated from artificial life.With virtual reality (VR) and intelligent agent techniques'emerging in the development of human-computer interaction (HCI), a novel agent-based VR HCI is developing under the guidance of life system. To make computer interaction more harmonious and friendlier, and to make computer have higher and comprehensive intelligence, we give computer the ability to adapt human emotion.This paper will introduce mainly the artificial emotion model of virtual human, and propose an improved genetic algorithm to estimate the model parameters. Combining the technology of facial expression recognition and virtual facial expression control, a virtual human system with emotion interaction was established.First, the establishment of emotion model can make computer be equipped with the basic ability to express and recognize emotion, and it has a wide range of applications in harmonious HCI. Emotion display can be divided into two layers: feeling and expression. In the course of human interaction, only the expression can be observed directly, the feeling is implied and can be only speculated or reflected through facial expression. It is similar to the characteristics of Hidden Markov Model (HMM), so an emotion model based on HMM was adopted in this paper.Second, in order to express the evolutionary process of the human emotion model more accurately, an improved genetic algorithm was proposed to estimate the model parameters. In order to overcome premature convergence of real-coded genetic algorithms, combining advantages of the two aspects of individual and population to propose a new crossover operator named heuristic weighted crossover operator.The third task is to establish virtual human system with emotion interaction. It takes static image as the system input, identify the facial expression of the image after the system's training, and then it calculates the type and intensity of the emotion according to the emotion model, then output the synthesized emotion of the system corresponding. In order to complete the emotion interaction, the synthesis technology of facial expression should be combined with the results presented before.Research shows that the virtual human system with emotion interaction is a harmonious interaction, and the emotion model is in accordance with the basic law of the human emotion.