期刊文献+

基于音频事件检测和分类的音频监控系统背景模型自适应方法研究 被引量:1

Research on Background Model Adaptation for Acoustic Event Detection and Classification Based on Acoustic Surveillance System
下载PDF
导出
摘要 随着监控系统中音频传感器应用的与日俱增,音频事件检测与分类已成为一个重要的研究课题。音频系统所处的音频环境(不同场所、不同噪声)非常复杂,以致检测与分类音频事件异常困难。因此,进行背景模型自适应从而适应不断变化的音频环境变得十分重要。提出了利用受限的最大似然线性回归方法对背景模型进行自适应。采用实际监控场景中的音频数据和模拟录制数据,研究了背景模型自适应方法以及如何有效地进行背景模型自适应。实验结果表明背景模型自适应可以提高目标声音事件的检测性能,减少系统误报。 Acoustic event detection and classification have become an important research problem as the increasing use of audio sensors in surveillance system. In these systems, audio circumstance is very complicated, that is, different loca- tions, different noises, which cause the acoustic event detection and classification to be very difficult. Therefore,it is im- portant to implement the background model adaptation in order to adapt these variations of background. In this paper, we proposed to use the constrained maximum likelihood linear regression (CMLLR) to adapt background model. Using the real world data and simulated data, we investigate the background model adaptation approaches and strategies for background model adaptation. Experimental results show that background model adaptation can improve the perfor- mance of target acoustic event detection and classification, and also can greatly reduce the false alarm of target acoustic event detection and classification.
出处 《计算机科学》 CSCD 北大核心 2016年第9期310-314,共5页 Computer Science
基金 国家自然科学基金(61305027) 山东省自然科学基金(ZR2011FQ024)资助
关键词 音频事件检测与分类 背景模型自适应 受限的最大似然线性回归 监控系统 Acoustic event detection and classification, Background model adaptation, Constrained maximum likelihoodlinear regression (CMLLR), Surveillance system
  • 相关文献

参考文献18

  • 1Wang D L,Brown G J. Computational Auditory Scene Analysis, Principles,Algorithms,and Applications [M]. Wiley-IEEE Press, 2006.
  • 2Espi M, Fujimoto M, Kinoshita K, et aL Feature Extraction Strategies in Deep Learning Based Acoustic Event Detection [C ]// INTERSPEECH. 2015 : 2922-2926.
  • 3Plinge A, Grzeszick R, Fink G A. A Bag-of-Features Approach to Acoustic Event Detection [C]///2014 IEEE International Conference on Acoustics, Speech and Signal Processing (IC- ASSP). 2014: 3704-3708.
  • 4Phan H, Maab M, Mazur R, et al. Random Regression Forests for Acoustic Event Detection and Classification [J]. IEEE/ACM Transactions on Audio, Speech, and Language Processing, 2015, 23( 1):20-31.
  • 5Parascandolo G, Huttunen H, Virtanen T. Recurrent Neural Networks for Polyphonic Sound Event Detection in Real Life Recordings [C]// 2016 IEEE International Conference on A- coustics, Speech and Signal Processing (ICASSP). 2016 : 6440- 6444.
  • 6Lim H, Kim M J, Kim H. Cross-Acoustic Transfer Learning for Sound Event Classification [M]//2016 IEEE International Con- ference on Acoustics, Speech and Signal Processing (ICASSP). 2016 : 2504-2508.
  • 7Atrey P K,Maddage N C,Kankanhalli M S. Audio based Event Detection for Multimedia Surveillance [C]//IEEE International Conference on Acoustics, Speech and Signal Processing (IC- ASSP). 2006 : 813-816.
  • 8Zhuang X,Zhou X, Hasegawa-Hohnson M, et al. Real-world A- coustic Event Detection [J].Pattern Recognition Letter, 2010, 31(12):1543-1551.
  • 9Zhang A Y. Using Hierarchical Method to Improve Real Time for Audio-based Surveillance System [C] /// International Sym- posim on Chinese Spoken Language Processing (ISCSLP). 2014: 570-573.
  • 10Rabaoul A, Davy M, Rossignol S, et al. Using One-Class SVM and Wavelets for Audio Surveillance [J]. IEEE Trans. on Infor- mation Forensics and Security, 2008,3(4):763-775.

引证文献1

二级引证文献1

相关作者

内容加载中请稍等...

相关机构

内容加载中请稍等...

相关主题

内容加载中请稍等...

浏览历史

内容加载中请稍等...
;
使用帮助 返回顶部