Research On Infrared Multi-camera Optical Tracking System For Surgical Robot

In the Robotics Assisted Surgical System, the position and orientation of surgical tools must be precisely and robustly estimated for the surgical navigation. In order to resist the variation of the visible-light in the work place and decrease the possibility of the occlusion of line-of-sight, an infrared multi-camera tracking system is developed. The optical tracking system involves the camera calibration; feature extraction; feature matching; stereo triangulation; tracking the 3D trajectories of multiple targets; device recognition and calibration.Our system uses a novel 3-D localization algorithm called Perpendicular Foot Method (PFM) and an automated Infrared-compatible calibration technique based on the Coordinate Measurement Machine (CMM). The experimental results indicate that this calibration accuracy is superior to the conventional method. Also, a novel multi-marker tracking algorithm using Extended Kalman Filter (EKF) is proposed, which makes full use of the redundant information of the multi-cameras. The target marker movement model can benefit from the multiple views by performing a special EKF updating 3D state by 2D projection observations on the image planes. It can estimate the movement of a 3D point more accurately by directly using the actual 2D observations from multiple views rather than error-prone 3D reconstructed point. This tracking algorithm is found to be able to track the 3D trajectories of multiple targets simultaneously, and the estimated 3D positions by EKF are in agreement with the actual measurements by the stereo triangulation named "Perpendicular Foot Method". Besides, we propose the geometric method for point pattern matching, and the method based on geometry is intelligibly and directly. Admittedly, this method cannot solve many other point pattern matching problems, but it is proved to be effective to recognize the rigid body. In addition to the innovative and applied techniques used in our system, we proposed some elementary theories and other various researches concerned the optical tracking system in this paper. In the second chapter, we describe the camera model and state some classic camera calibration methods. The 3D reconstruction theory including feature matching and stereo triangulation is presented in the third chapter. We descript the Extended Kalman Filter simply and its application in our system detailedly in the chapter four. The fifth chapter proposes an actual problem concerned our appliance, how to recognize and calibrate the surgical device. Overall, this dissertation is an exhaustive report involved the multi-camera optical tracking system used in Robotics Assisted Surgical System.