Visual Projection Interaction System Based On Linear Classifier

Augmented Reality(AR) is the integration of virtual object with the user’s environment in real time, which allow the users interact with the virtual environment. AR creates a new interactive experience for the users and give them immersive feeling. Interaction and visual projection system is an important application of AR. It relies on advanced computer vision and projection display technology, which can build a kind of more practical experience interactive mode. In this paper, discussion is focused on the following three aspects.(1) Firstly, after introducing the background and current research status of Interaction and visual projection system all over the world, this paper puts forward a new solution, based on linear classifier technique, for visual projection interaction system. This solution using infrared light to shine the target, and then designs two stage classifier to identify the imaging of the target. In addition, to provide theory basis for the design of this system, this paper also introduce the relative theory of interactio n system in detail.(2) Secondly, according to the visual projection interaction system’s framework, this paper design a set of hardware and software process of the visual projection interaction system. For the hardware platform,it includes an USB camera, a HD portable projector and a linear infrared light. For the software platform, it includes two systems. O ne of them is a total process software system, including image acquisition and correction, image preprocessing, target capture, target classifier and the target behavior recognition. The other one is a desktop software, which for learning and training the secondary linear classifier.(3) Thirdly, with the finished hardware platform, this paper have done each modules of the above software system step by step. And then, getting a large number of statistical data of three key modules for analysis. The three modules include image correction, first classifier for total process software system and the secondary classifier based on pattern recognition. Finally, the statistical data demonstrate that the system is working well.