A study on surgical robot image stabilization

This study aims to investigate and analyze various image stabilization methods used in surgical robotics. An in-vitro phantom experiment was conducted using a master slave tele-manipulated system. An articulable image probe on a tool (eye-on-tool) was installed on an open-source surgical robot, RAVEN. The 2D non-stereo camera image was used to validate the image stabilization methods, which were evaluated individually for each processing step: preprocessing, motion estimation, and motion compensation. During the preprocessing procedure, the performance of three different filters was tested for effective noise suppression. Various algorithms were compared to estimate the global motion vectors (GMVs) in the motion estimation step. Finally, three filters were analyzed to estimate the compensation motion vector (CMV) during the motion estimation procedure.