An efficient method for video summarization using moving object information

TL;DR: A novel video summarization scheme using moving object information by considering area of moving objects extracted from dynamic background modeling and frame-to-frame object motion to rank all frames according to the importance of being key frame by combining moving object features through a fusion method.

Abstract: Video surveillance system captures continuous video for the purpose of security, monitoring, investigating and so on. It requires huge memory space to store as well as enormous time to retrieve important information manually from this high volume of videos. In this paper, we propose a novel video summarization scheme using moving object information by considering area of moving objects extracted from dynamic background modeling and frame-to-frame object motion. Through this scheme we rank all frames according to the importance of being key frame by combining moving object features through a fusion method so that users can select desired length of videos for summary. The experimental results show that the proposed method provides better video summary compared to the state-of-the-art method using a publicly available benchmark BL-7F video surveillance dataset.