文章导读
总览 评价 莫能斌 , 刘刚 * ( 北京邮电大学信息与通信工程学院,北京 100876; ) 摘要: 传统的安全监控是基于视频图像的,而且大部分主要用于事后查看。近年来,音频监控也得到了许多研究。音频监控因其良好的实时性以及对视频监控的辅助,具有重要的研
莫能斌, 刘刚*
(
北京邮电大学信息与通信工程学院,北京 100876; )
摘要:
传统的安全监控是基于视频图像的,而且大部分主要用于事后查看。近年来,音频监控也得到了许多研究。音频监控因其良好的实时性以及对视频监控的辅助,具有重要的研究和实用价值。音频监控是通过音频事件检测来进行的,本文针对安全监控下的四种音频事件做检测,包括脚步声、玻璃破碎声、开关门声和语音。首先采用MFCC特征搭建最初的音频事件检测系统,之后采用两种平滑方法对结果做平滑,并提出一种新的平滑方法,结果显示新方法效果良好。最后针对训练数据不平衡问题,做了一些随机欠采样的实验。总的结果表明音频事件检测系统的准确率不高,但是召回率达到了比较好的效果。
关键词:
模式识别;音频事件检测;安全监控
Mo Nengbin, Liu Gang*
(
School of Information and Communication Engineering, Beijing University of Posts and Telecommunication, Beijing 100876; )
Abstract:
Traditional safety monitoring system is based on video, and mostly the video is reviewed afterward. In recent years, audio surveillance has been researched by many researchers. Because of its nice real-time performance and the auxiliary effects to video monitoring, audio surveillance has important research and application value. Audio surveillance is based on audio event detection. In this paper, we intend to detect four audio event categories, including footsteps, sound of breaking glasses, sound of opening or closing door and speech. Firstly we choose MFCC feature to construct the initial audio event detection system, then we proceed two kinds of smoothing method. Afterwards, we propose a new smooothing method and get good result. To solve the problem of unbanced training data, we perform some down-sampling experiments. The total results show that the accuracy of the system is not very good, but recall rate is rather good.
Tag:
点此返回栏目查看更多>>>参考论文