Research On Registration And Occlusion Based On Kinect

Augmented reality consists of computer graphics, image processing and computer vision?The ultimate purpose of augmented reality technique is to integrate the virtual environment with actual environment and augmented reality for providding Additional information to explain the real word. as a kind of body sensor,Kinect with augmented reality technology will provide more interesting applications. Real-time registration and the actual situation of shade implementation is the key to the Kinec augmented reality technology, but most of the augmented reality system can not solve the problem.Through the analysis of the problems existing in the augmented reality, we build a augmented reality system based on 3d reconstruction. The system make full use of color image and depth images at the same time, and it use the cloud point matching algorithm and optical flow method for the camera pose tracking.The system can track camera real-time in the extended reconstruction area. By using the result of the two camera calibration and Reconstruction information, the system of virtual model and real scene images have more accurate occlusion relations.Experimental results show that based on the reconstruction combining iterative closest point matching algorithm and optical flow algorithmmake the tracking more stable than only using iterative closest point matching algorithm.By extending the scope of reconstruction to the original scope of 2 to 3 times, augmented reality scenarios scope also expanded. Using the depth of the scene scene information, virtual model and the accuracy of the real scene sheltered relations is accurate.