During laparoscopic procedures, the main concern for surgeons is to take the most appropriate decisions on the basis of a fragmented view of the operating field. This task is quite difficult to achieve since the endoscope offers a restricted field of view of the abdominal cavity and is constrained by the insertion point. These constraints lead to constant back and forth movements of the endoscope, swapping between close detailed views and more global ones in order to locate anatomical landmarks. Moreover, surgeons would often like to radically change the viewpoint, in order to be able to observe partially hidden anatomical parts, without inserting a new trocar. Our purpose is to provide the surgeon with a system that instantaneously creates a global and detailed view of the operating field from any desired viewpoint. This could significantly enhance the medical service offered by laparoscopy. In order to do so, we insert a commando of n small CMOS video cameras in the patients abdomen through standard endoscopic trocars. Some of these cameras can be grouped into stereoscopic pairs. They are attached to trocars, to surgical instruments or to strategic parts of the abdominal wall, so that at least one of them is perfectly positioned to capture the relevant visual clues. A 3D model of the operating field is created from all the partial fields of views of the n cameras. Geometrical models and color informations are merged to synthesize any virtual view of the abdominal scene. The surgeon may thus explore virtually around the observed organs without moving any real camera.
