Summarizing Videos by Key frame extraction using SSIM and other Visual Features

Video Summarization is the generation of summaries of videos of any domain. The summaries are compact representation of videos and consist of important events that occur in a video. Video Summarization gives flexibility to users to understand the content of video without watching the whole video itself. In this paper, video summaries are generated by extracting key frames of a video using visual features like SSIM, color histogram, moment measure and correlation measure. The summaries are similar to human perception. The technique also works well for videos of varying illumination. The technique is tested on data set available at OpenVideoProject. The videos taken for experiments are Ocean Floor legacy, 25th anniversary of NASA segment 04, anni9. avi and The Voyage of the Lee segment 05. All the videos are of duration less than 10 minutes. The proposed technique is compared with existing techniques like VSUMM and STIMO.