Algorithm for Stabilizing Fast Forward of Videos, Especially for Wearable Cameras
Cluster5 Project ID : 31-2015-4224
Summary of the technology
Project ID : 31-2015-4224
Description of the technology
Computer Science & Engineering, Imaging / Computer Graphics
Proof of concept
Patent application filed in the United States
The videos captured by such egocentric cameras such as GoPro and Google Glass are long, boring, and difficult to watch from start to end.
Attempting to browse the video faster by fast forwarding based on frame sampling is problematic in egocentric videos because the shake introduced by natural head motion gets accentuated rendering the video useless.
There is thus a need for automated tools that enable faster access to the information in such videos.
This solution provides a novel and lightweight approach for creating fast forward videos for egocentric videos and can be used as a method to create stereo sequences from monocular egocentric video.
Ego-Sampling, a novel frame sampling technique to produce stable fast forward for egocentric videos
Representative frames from a fast forward from a camera wearer riding a bike and preparing to cross a road. Top row: uniform sampling of the input sequence leads to a very shaky output as the camera wearer turns his head sharply to the left and right before crossing the road. Bottom row: Ego-Sampling prefers forward looking frames and therefore samples the frames non-uniformly so as to remove the sharp head motions. The stabilization can be visually compared by focusing on the change in position of the building (circled in yellow) appearing in the scene. The building does not even show up in two frames of the uniform sampling approach, indicating the extreme shake.