| Because of the laparoscope's high magnification and narrow field of view, navigation and visualization in laparoscopic surgery is difficult. An image processing system is proposed to manipulate image data from the laparoscope in real-time to improve visualization during laparoscopic surgery. The system maintains "visual momentum" by broadening the field of view and smoothing fast viewpoint transitions. Consequently, the surgeon can navigate and visualize more easily in laparoscopic surgery. A prototype system is developed to test the concept. The field of view is broadened by real-time image mosaicing. Fast viewpoint transitions are smoothed by real-time view interpolation. The foundation of the mosaicing and interpolation algorithms is image registration, the mapping of images taken at different viewpoints. The theory, implementation, and results of the system are presented and discussed. A new featureless image registration algorithm is also proposed to solve the problem of lack of features in laparoscopic images. |
