A Surgical Bullet-Time Video Capturing System Depending on Surgical Situation

Yinghao Wang, Chun Xie, Hidehiko Shishido, Shinji Hashimoto, Tatsuya Oda, Itaru Kitahara

Abstract: In recent years, capturing surgical video has become popular for education and case studies in surgical technology. However, there is always a problem of obscuration caused by the surgeon's head or hands when using a fixed surgical camera for surgery. To improve the quality of surgical videos, we propose a capturing system with multiple cameras installed between the surgeon and the patient. In this paper, we propose an effective method to automatically switch the viewpoint of multi-view surgical videos. When there is an occlusion in the current view, the system can switch to other cameras and rearrange the captured video sequences into bullet-time videos to ensure that details and surgical procedures are recorded. We also adopt a learning-based object detection method to automatically set the gaze points of certain objects from the multi-view surgical video.