Research Of 3D Registration Algorithm Based On Particle Filter

Augmented Reality (AR) is an important branch of the development of Virtual Reality. With scientific and technological development,the technology of augmented reality is constantly improved. It promoted AR in medical, manufacturing, military, entertainment and other aspects of the application. AR is a technology in which a computer-generated image is superimposed onto user's vision in the real world, giving user additional information generated from the computer model. Because of its application in the future, AR has become an important research direction of computer science.Because domestic research on AR technology starts late, most of studies also remain in the application. There are fewer studies on 3D registration which is the key technology of AR. Most research on 3D registration focus on bottom-up approach for which the matrix of 3D position and 3D orientation is calculated from low-level 2D features and their 3D geometric relation. In contrast to bottom-up approach, this paper introduce particle filer algorithm to 3D registration. The matrix is obtained from top-down state space approach using motion model and prediction.This paper establishes the system model of 3D registration based on particle filter and described in detail the process of the algorithm. On this basis, we implement a software system based on our algorithm. Based on detailed analysis of experimental results, we show that in AR application the algorithm is adaptive and robust. The paper further proposes that an adaptation method for the particle filter. The experimental results show that the method is more stable than previous methods.