摘要
针对现有驾驶疲劳状态识别算法中存在疲劳特征维数高、识别效率低下、计算量大等问题,本文提出一种基于在线字典学习形变模型的疲劳状态识别方法。采用红外疲劳人脸图像中关键变形区域LBP特征构建人脸形变模型;将在线字典学习算法引入到形变模型中,采用过完备基函数矩阵代替训练样本整体对待测样本进行线性表示,利用其组合系数的稀疏性进行人脸疲劳状态识别;采用时间窗结合贝叶斯方法对识别算法进行优化。实验结果表明,与传统的识别方法相比,本文所提算法可以降低系统的运算量,提高疲劳状态识别的鲁棒性和准确率,在实际驾驶环境中能够取得良好的识别效果。
Some problems such as high feature dimension,low recognition efficiency,and large amount of computation are drawbacks in existing driving fatigue state recognition algorithms. To tackle these problems,a new fatigue state recognition method based on the online dictionary learning deformation model was proposed. First,the LBP features of a key deformation region in infrared fatigue face image were applied to construct the deformation model.Second,the online dictionary learning algorithm was introduced into the deformation model,in which the over-complete base function matrix was employed instead of the whole training sample to express the test face images. The sparsity of linear combination coefficients was used to recognize the state of face. Finally,the time window and Bayesian theory were combined to optimize the recognition algorithm. Compared with traditional recognition methods,the proposed method improves the robustness and recognition rate of the system and reduces the computational complexity of the system. Therefore,the proposed method can obtain good recognition effects in an actual driving environment.
出处
《哈尔滨工程大学学报》
EI
CAS
CSCD
北大核心
2017年第6期892-897,共6页
Journal of Harbin Engineering University
基金
国家自然科学基金项目(51479042)
关键词
疲劳状态识别
变形区域
LBP特征
形变模型
在线字典学习
过完备基函数矩阵
时间窗
贝叶斯方法
fatigue state recognition
deformation region
LBP features
deformation model
online dictionary learning
over-complete basis function matrix
time window
Bayesian theory
