摘要
为了解决传统接触式疲劳驾驶检测方法影响驾驶、检测算法识别率较低等问题,本文提出一种基于稀疏表示的眼睛状态识别的方法。利用K-SVD(K均值奇异值分解)方法对输入的训练集构造过完备冗余字典,利用正交匹配追踪法对测试的图像进行稀疏表示,然后根据重构图像和测试图像之间的误差,确定测试图像所属的类别,判断出测试图像的状态。实验中将KSVD和OMP(正交匹配追踪)方法与其它字典学习和稀疏表示方法进行对比,结果表明,利用K-SVD字典学习算法结合OMP算法获得了较好的识别效果。
In this paper,we proposed an eyes state in driver fatigue recognition algorithm based on sparse representation in order to solve the problems that in traditional contact detection,just like affect driver,low recognition rate,etc.First,construct a complete redundant dictionary to training set using K-SVD method and sparse representation the test images employing orthogonal matching pursuit method.Then determine the error between the test images and the reconstructed images.At last,determine the type of test images and judge the state of test images.And contrast the K-SVD to the other dictionary learning methods and sparse representation methods,the experimental results showed that the K-SVD algorithm combined with OMP obtained better recognition effect.
出处
《影像科学与光化学》
CAS
CSCD
北大核心
2016年第1期95-101,共7页
Imaging Science and Photochemistry
基金
吉林省科技发展计划项目(20120434)资助
