Domain-Aware Video Summarization with Motion Information and Foreground Object Fusion

MSRDG International Journal of Computer Scientific Technology & Electronics Engineering

 

© 2025 by MSRDG IJCSTEE Journal

Volume 1 Issue 1

Year of Publication: 2025



Authors: S. Alangarom, A. Vijay
Paper


Download


Article ID
MSRDG-IJCSTEE-V1I1P105
Abstract:

Surveillance cameras produce tremendous amounts of continuous video data every day, which is hard and time consuming to determine important events by hand. In this paper, the authors propose a “video summarization method based on foreground” object detection and motion information from spatial and frequency domains to solve this issue. The approach models the background and identifies motion clues to extract foreground objects successfully. “Motion in the spatial domain” is obtained from frame changes, whereas ”frequency domain motion” is obtained through the Phase Correlation (PC) method. Through the combination of motion information and foreground object data in both domains, the system detects and retrieves key frames optimally representing the content of a video. Experimental results validate that the suggested method performs better than other approaches in effectively and accurately summarizing surveillance videos.

Keywords: Surveillance video, Video summarizing, Spatial and frequency domains, Foreground objects