Home > About Fuji Xerox > Technology > New Technologies for Valuable Communication > Multimedia Processing Technology > Video Keyframe Extraction Technology

Video Keyframe Extraction Technology

In line with the growing popularization of digital cameras and smartphones, people are capturing more video than ever before. This practice results in more videos being stored on storage devices. In order to find a specific scene in a video file from such a large collection of videos, one must play each video and watch its content. Thus, the time spent on finding a desired scene increases as the number of stored video files grows. Fuji Xerox has hence developed Video Keyframe Extraction technology that selects images representing the content of a video from all its constituent frames, and identifies these images as keyframes. With this technology, we can quickly browse the content of a video without playing it.


Fig. 1: An example of extracting keyframes by evaluating the difference in visual features between consecutive frames


Fig. 2: An example of extracting keyframes from clusters formed based on visual similarity without consideration of temporal order

Related information

Case example