Researches On Tracking The Boundary Of People And The Basic Implement Of The Advanced User Interfaces

The Advanced User Interfaces is an important branch of the Human's Motion Analysis in the application domain. The Human's Motion Analysis is to detect, recognize and track the people from the image sequence and understand the people's behavior. The Advanced User Interfaces based on vision studies the interaction between the computer and people. It has achieved great success in the business and application domain.The Advanced User Interfaces will involve widely and include the fields such as motion detection, object classification, people's tracking, gait recognition, voice recognition and behavior understanding. When we apply it for business and application system development, more fields will be involved such as architecture design and algorithm design. In this thesis, we will emphasis on the following two areas.First, theorectically, we will dig into the theory of boundary tracking and propose our new algorithm. Boundary tracking is an important part in the Advanced User Interfaces system. The tracking algorithm using the level sets has received lots of attention in recent years. In this thesis, we proposed a novel algorithm based on the dynamic boundary feature for tracking the boundary of a moving people using the level sets.Second, in the application domain, we will develop a real but simple application which can be regarded as a basic implemention of the Advanced User Interfaces. It is a video-interaction game which is something like the famous Sony entertainment game - EyeToy. We develop our game based on the open source project OpenCV. This game will include the many fields about The Advanced User Interfaces such as tracking of moving objects, understanding and reaction to the man's behavior.Experimental results related to the efficiency and accuracy of these proposed methods are also presented in the thesis.