Abstract: Traditional brute-force stereo matching techniques that use local match measures are time consuming and often have limited accuracy. In this talk, I will present a surface tracking algorithm that considers the correspondence problem as a time-varying optimization of a parametric function. By reducing the computational complexity, this method is able to provide fast and accurate 3-D visualization of the image scenes. The ability to process images online allows the algorithm to be used in a number of surgical applications, including establishing safety regions, developing virtual fixtures, and measuring mechanical properties of various tissues and organs. In my master thesis work, I investigated the potential of this algorithm as part of a virtual stabilization system for compensating residual motion of the heart during robot-assisted off-pump coronary artery bypass surgery (CABG). In this system, the robotic arms will move in sync with the computed heart motion, creating a virtually stable surgical field. In this presentation, I will demonstrate the results obtained from processing image sequences of beating pig hearts. The clinical significance of this work and potential for future work will also be discussed.