Pose Estimation And Image-Based Rendering Application In Computer Graphics On Android Platform

Through the motion of the camera, the computer can get the real-time 2D image, so that the motion parameters of the camera can be obtained by the correlation algorithm. The analysis of camera motion parameters belongs to the category of computer vision, and the 3D model loading belongs to computer graphics. The application of camera motion in computer graphics is a cross research project of computer vision and computer graphics. Typical research topics are augmented reality (Reality Augmented), virtual reality (Reality Virtual). Augmented reality through the input of sensors, such as sound, video, or GPS sensors to generate direct or indirect enhancements to the physical, real world, through the computer. Augmented reality is to strengthen the real world with the virtual world, while the opposite is true, virtual reality by replacing the real world for the computer generated virtual world. Augmented reality using computer vision technology based on camera motion, such as three-dimensional reconstruction, scene tracking and other techniques based on motion. And virtual reality through a variety of motion sensors, such as the gyroscope, GPS tracker to obtain motion information, the device’s motion to the camera movement, and reflected in the virtual world.This paper makes a brief introduction on the technology of augmented reality and virtual reality field, including the theoretical basis of the multiple view geometry, the basic principle of 3D modeling, based on image rendering technology are described. The realization of this paper is based on the application of augmented reality based face recognition. Using the efficient face detection algorithm. By off-line on the face to do 3D face model, the posit algorithm based on 2D face feature point and 3D model vertices correspondence get camera motion matrices to compute the face pose. And in the Android mobile phone to achieve augmented reality effect to observe the effect of human face pose detection. The experimental results show that not only the facial pose estimation effect is obvious, but also the real-time effect of the 3D face model is achieved. In this paper, the realization of the second system is based on virtual reality three-dimensional virtual display system, by off-line generation of camera motion, and fitting regular trajectories, through two adjacent view camera parameters, based on three view geometry of pole shift principle generated along the camera trajectory at any point of view of the camera parameters perturbation. Then based on Algorithm with preserving the integrity of the content is generated along the new photo camera trajectory on any view. The experimental results show that compared with other image based rendering algorithm, such as panorama, Photosynth, view morphing, algorithm, not only to maintain the content, and expand the scope of view generation.