Online semantic condense video in robotic scene via dictional selection
Lists
Only you see this
Abstract
Video data are important representatives for information storage. However, high similarity always exists among the video frames in many cases, such as robot videos, some kind of surveillance videos, and so on. This will lead to information redundancy, and futher will cause problems in information storage, query, and transfer. In this paper, we present an algorithm for video semantic enrichment. This method may perceive the video content quickly and accurately by extracting the key frames from video sequences. First, we extract the continuous foreground segments from the original video sequences to remove the background frames. Then, we segment the foreground to be a series of sub video sequences by the shot boundary detection process. The features extracted from the sub video sequences constitute the original dectionary. Finally, we select the key frames with a dictionary selection process. By removing the useless information with a background model, our method can deal with arbitrary length videos by the shot boundary detection. The meaningful key frames can be extracted based on the image features and dictionary seletion method. Compared with the artificial identification method and the baseline algorithms, the experiment results verify that our method can obtain more accurate key frames.
BibTeX
@inproceedings{cong2013online,
author = {Yang Cong and Hongwei Gao and Jiangping Huang and Haibin Yu and Yandong Tang},
title = {{Online semantic condense video in robotic scene via dictional selection}},
booktitle = {Chinese},
year = {2013},
doi = {10.1360/972013-951},
}Loading related content...