摘要: 分析目前应用于背景提取的各类聚类方法的原理和存在的问题,提出一种基于自适应在线聚类的背景提取方法。通过使用自适应动态改变的聚类阈值对视频进行在线聚类,无须设定任何参数即能自适应地提取出背景图像。实验结果表明,该方法具有较好的自适应性,能够提取出较优的背景图像,对于各种视频具有较好的鲁棒性。
关键词:
背景提取,
K-均值聚类,
在线聚类
Abstract: By analyzing the principles of various clustering methods and existing problems when current clustering methods are used in background extraction, this paper proposes a background extraction method based on adaptive on-line clustering. This method adopts adaptively dynamic thresholds to do on-line clustering on video to adaptively extract background images from a variety of videos without setting any parameters. Experimental results show that this method has a good adaptability, and is capable of extracting excellent background image, and has good robustness for various videos.
Key words:
background extraction,
K-Means clustering,
on-line clustering
中图分类号:
夏洁, 吴健, 陈建明, 崔志明. 基于自适应在线聚类的背景提取[J]. 计算机工程, 2011, 37(3): 169-171.
JIA Ji, TUN Jian, CHEN Jian-Meng, CUI Zhi-Meng. Background Extraction Based on Adaptive On-line Clustering[J]. Computer Engineering, 2011, 37(3): 169-171.