To convert videos based on a plurality of video data generated at a plurality of imaging devices having different angles of an optical axis respectively to make a user perceive the videos as a natural stereoscopic video without a sense of discomfort.

A data acquisition part 110 of a stereoscopic video processing device 100 acquires left-eye video data and right-eye video data. An imaging state information acquisition part 112 acquires imaging state information indicating at least two or more imaging states among a baseline length, a congestion point distance, and a congestion angle. Based on the imaging state information, a video conversion part 114 performs video conversion of a left-eye video based on the left-eye video data so as to map the left-eye video onto a display surface 108 from a left-eye virtual surface 152a orthogonal to an optical axis 106a of an imaging device 102a that generated the left-eye video data, and performs video conversion of a right-eye video based on the right-eye video data so as to map the right-eye video onto the display surface from a right-eye virtual surface 152b orthogonal to an optical axis 106b of an imaging device 102b that generated the right-eye video data.