The Research On Key Technology Of Augmented Reality Based On Visualization

Augmented Reality(AR)is a view of a physical real-world environment whose elements are augmented by computer-generated virtual objects,virtual scene and system information depending on display,human-computer interaction and computer graphics and multimedia technology.It makes the users feel that the virtual information generated by the computer is just a part of the real world around them.The key technology of AR mainly includes registering,virtual-real fusion,lighting,displaying techniques and interaction techniques.All these techniques are aimed at assuring the virtual information added in real world to be accurate so that users feel the system more real and immersive.To further improve the robustness and accuracy of AR system,the key technologies of visual object tracking,3D registration and virtual-real interaction are systematically and deeply researched in this paper.The main contributions of this paper are as follows: 1.Aiming at the problem that the fast tracking algorithm using spatio-temporal context will inevitably lead the drift and even lose the target in long-term tracking,a long-term tracking algorithm based on spatio-temporal context that integrates long-term tracking with detecting is proposed in this paper.First,we track the target by the fast tracking algorithm.Then the cascaded search strategy is introduced to the detecting part to relocate the target if the fast tracking fails.Meanwhile,according to the spatio-temporal correlation in video sequence,the predictive motion vector is added to searching algorithms so as to increase the detecting speed.To a large extent,the proposed algorithm effectively improves the accuracy and stability of long-term tracking.Extensive experimental results on benchmark datasets show that the proposed algorithm can accurately track and relocate the target though the target is partially or completely occluded or reappears after being out of the scene.2.Aiming at the problem of mark-less 3D registration algorithm has high complexity and difficulties of detecting feature points when the motion blur occurs,a mark-less register method based on visualization was proposed.Before registration,the proposed long-term tracking algorithm based on spatio-temporal context is used to track the register target so as to improve the operating speed and accuracy of detecting feature points.Meanwhile,to meet the requirement of accuracy and stability of detecting feature points under both situation of motionless and motion blur,different characteristics of detecting feature points of ORB algorithm and KAZE algorithm are used.ORB algorithm has the rapidity as it is based on FAST algorithm and Brief algorithm.And KAZE algorithm can effectively keep image edge by non-linear scale space when motion blur occurs.Finally,RANSAC algorithm is used to calculate homography and then the registration is completed.The experiment results show that under different condition the proposed method can accurately detect the feature points,register and has good robustness in illumination,occlusion and motion blur.3.A new method of 3D gesture interaction based on Leap Motion is proposed in AR virtual-real interaction.First,the mark-less registration algorithm based on visualization is used to register the virtual model.Considering that interacting using traditional data glove is inconvenient,Leap Motion controller is used to track and recognize the gesture of user.Additionally,single-hand and both hands gestures are designed to control the 3D model in six degrees of freedom.The threshold and SVM method are used to recognize user's gestures.Finally,the vessel model is taken as the interaction instance to do surgery simulation which testifies the effectiveness and naturalness of virtual-real interaction and to some extend improves the application of AR on medical practice.To make the AR system more accurate and robust,visual object tracking,3D registration and virtual-real interaction are systematically and deeply researched in this paper.Extensive experimental results shows that the proposed long-term tracking algorithm based on spatio-temporal context improve the accuracy and stability by fusing tracking and detecting.Additionally,the proposed mark-less registration algorithm can accurately detect the feature points,register and has good robustness under different conditions.Finally,the user can naturally and effectively interact with the virtual model by designing and realizing the virtual-real interaction method based on Leap Motion,which further improves user's sense of immersion in AR system.