摘要
疲劳是造成交通事故的主因之一,提出了一种基于Gabor小波变换的疲劳监控新方法。首先,在训练阶段采用频繁模式挖掘算法对疲劳脸部图像序列集进行疲劳模式挖掘;然后,在疲劳识别阶段,将待检测的脸部图像序列基于Gabor小波变换表示为融合特征序列;最后,采用分类算法进行人脸序列的疲劳检测。对自行收集的一天内500幅疲劳图像的仿真结果表明,所提方法正确检测率达到92.8%,错误检测率达到0.02%,优于比较算法。
Fatigue is one of the main factors that cause traffic accidents.A new method for monitoring fatigue state based on Gabor wavelet transform was proposed.In this method,the frequent patterns mining algorithm was designed to mine the fatigue patterns of fatigue facial image sequences during the training phase first.And then,during the fatigue recognition phase,the face image sequence to be detected was represented by fused feature sequence through Gabor wavelet transform.Afterwards,the classification algorithm was used for fatigue detection of the human face sequence.The simulation results on 500 fatigue images sampled by the authors show that the proposed algorithm achieves 92.8% in right detection rate and 0.02% in error detection rate,and outperforms than some similar method.
出处
《计算机应用》
CSCD
北大核心
2011年第8期2119-2122,共4页
journal of Computer Applications
