The Research And Development Of Actual-real Registration Based On Feature Points

Augmented reality has become a very popular area of research because of the progress of the computer technology. Augmented reality technology will make the real images from the camera and the computer generated virtual image synthetic accuracy and seamless and illustrate through the camera, helmets, displays, etc. and realize man-machine interaction in real time. And the key technology to construct a successful augmented reality system is the real images and virtual image synthetic accurately. This topic is the research of the virtual-real registration technology, and a remote Chinese chess game system is designed to realize the technology. The research results can be applied to artifacts at the museum of cultural relic’s protection and recovery, and remote medical and remote education and remote entertainment and so on.The main researches are as follows:1) For an existing feature point extraction algorithm extracts the slow, low noise performance and scaling on the scales sensitive issue, feature extraction algorithm based on improved SIFT descriptor operator is studied and proposed. The algorithm for analysis on the basis of experiments of the existing feature point extraction algorithm and the chess board, can quickly identify the four corners of the board, and further identify the other line and the intersection location. Experiments show that feature extraction algorithm based on improved SIFT descriptor operator is feasible and effective.2) For the traditional Mean Shift algorithm the phenomenon of the target with the lost in the case of high similarity prone of the target and the background, making full use of the feature point extraction technique based on the results of the research, a SSIFT-Mean Shift algorithm is studied and proposed. The algorithm uses SSIFT algorithm extracts the defined target area of the corner, and then builds track candidate regions; the tracking results of each frame is showed. Experiments show that this SSIFT-Mean Shift feature point tracking algorithm is feasible and effective.3) For remote chess play system, accuracy and real-time real and virtual registration technology is raised up, which implements a remote chess game system this side pieces accurately overlay virtual networks pass over from the other side of the board.4) Remote chess play system based on augmented reality is designed and implemented. In this paper, seven important module is designed and implemented of three-dimensional video capture, board identification, chess motion detection, network transmission, the actual situation registration and strength synthetic, three-dimensional model processing and three-dimensional video output, the final completion of the system. Remote Chess game system provides users an immersive feel, in addition to the thrill of a game, but also feels at home.