| With the technology advance and hardware improvement in the information era, it's an inevitable trend that traditional styles of communication give way to new visible styles. At the same time, the privacy problem which appears when we transfer visual images with each other comes to the front. On the other hand, expression recognition, one of the important research fields in computer science, has become the basic element to make computer understand the feeling and intention of humans in the user-computer relation.The compression and coding of facial image are one of the hot topics in the field of digital image processing techniques widely used in communication and internet. The problem is divided into three steps:accurate face locating algorithm, real-time face tracking algorithm and efficient face coding algorithm.This paper is mainly about the process of developing the expression simulation plug-in (facial expression recognition system) which includes need analysis, application field, design intention and most importantly, the theory and algorithm and the core codes of the project.The principle innovation of this paper is that, while the traditional way of cutting face by the skin color is greatly depend on the center of statistics of skin color distribution, we give an innovation algorithm by cutting twice and adjust the center of skin color distribution after the first cut to make it closer to the actual center. Also in this paper in order to accelerate the computing speed, the pixel color characteristic vector and the Euclid distance on distribute center are adopted to avoid the complex calculation of similarity.The project uses related cartoon expressions instead of actual facial expressions to protect the privacy of the user making the communication available at the same time and it decreases the load of internet by just transferring the characteristic vector instead of the whole images. All in all, the paper has its value both in application and in innovation. |
PDF Full Text Request