Automated Motion Feature-Based Video Synchronization

A system includes a hardware processor and a system memory storing software code. The hardware processor is configured to execute the software code to receive a plurality of video sequences each generated by a respective one of a plurality of cameras, each of the video sequences depicting a respective one of a plurality of three-dimensional (3D) movements of a subject from a different perspective, extract, from each of the video sequences, a respective set of two-dimensional (2D) motion features, to provide a plurality of 2D motion feature sets, and synchronize, using the 2D motion feature sets, the plurality of video sequences.

BACKGROUND

In applications in which multiple video cameras are used to record an event, it is often necessary to synchronize the videos in order to sensibly merge the different perspective each camera captures. For example, different smartphone camera videos of the same event, such as a sporting event, a concert, a show, a parade, or an emergency situation for example, may be recorded by different witnesses to the event. Moreover, such “witness videos” are often initiated, terminated, or both, at different times by the respective attendees or witnesses.

Conventional approaches to merging videos typically require that those videos already be synchronized, rely on manual synchronization techniques, or require the use of expensive synchronized cameras. These conventional approaches rely on features or technologies that are often not germane to the underlying event that is the subject of the videography, and thereby undesirably introduce the use of unnecessary and extraneous elements. Consequently, there is a need in the art for a solution enabling the automated synchronization of videos based solely on movements by the subjects depicted in those videos.