UC San Diego

Over the past several years, a dramatic increase in attention has been paid to 3D media, particularly in the movie industry. Due to the lack of previous algorithmic advancement and the absence of unobtrusive technology, however, three-dimensional visualization has failed to penetrate the operating room until now. In this work, we present a system to deliver glasses-free 3D visualization of laparoscopic surgeries to the operating room via multiview autostereoscopic displays. We begin by developing robust stereo-to-multiview content generation so that we may produce an arbitrary number of stereo sequences for presentation on autostereoscopic displays. We then introduce a reliable disparity estimation technique that enables the synthesis of the virtual views. In addition, we enforce spatio-temporal consistency in disparity estimates to provide a superior 3D experience for surgeons and operating room staff. Finally, we implement the entire system on graphics hardware in order to achieve real-time operability. Numerous attempts have been made to introduce three-dimensional (3D) video systems into clinical routine, particularly for surgeries. The drawback with all of them thus far has been the fact that they require users to wear cumbersome glasses in order to receive the advantage of stereoscopy. In this work, we present, to our best knowledge, the world's first laparoscopic surgical system that delivers glasses-free multiview 3D in high-denition (HD) resolution. In addition to a quantitative evaluation of the video processing components, we performed an initial subjective study with laparoscopically experienced surgeons, which yielded very promising results