With the increasing popularity of laparoscopic surgery, the demand for better modes of laparoscopic surgery is also increasing. The current laparoscopic surgery mode requires an assistant to hold and manipulate the endoscope through commands from the surgeon. However, during lengthy surgery procedures, accurate and timely adjustment of the camera cannot be guaranteed due to the fatigue and hand trembling of the camera assistant. Another common problem in laparoscopic surgery is its lack of 3D depth cues. The surgeon has to develop a two-dimensional concept for manipulating tissues and moving instruments across a flat video screen.

This thesis proposes a novel and practical image tracking method to achieve automated instrument localization and endoscope maneuvering using a solo endoscope in laparoscopic surgery. Aspects of this approach, such as camera calibration, marker design, endoscopic image distortion correction, endoscope manipulator design and simulation, and image tracking are described in detail. In our work, we propose to attach a designed marker to the instruments. In the tracking process, the instrument's x and y positions are obtained by locating the center of the marker. The depth information, z, of the instrument is calculated from two points on the boundary of the marker which represent the diameter of the instrument. Thus, 3D tracking is realized by using only one endoscope. Experimental results show the feasibility of the proposed methods. When compared with another position sensing system, our tracking results provide better accuracy. To deal with the lack of 3D depth cues in laparoscopic surgery, an active triangulation method is proposed to obtain range data from the scanned internal surface. We then reconstruct 3D surface images from range data. This thesis presents the basic design for obtaining range data and 3D surface reconstruction. Experiments show that the laser range camera can achieve accuracy within 1mm and the reconstructed surface closely represents the real object (the reconstructed slope angle error is within 0.5°).