摘要
针对人脸图像识别精度不高的缺点,本文将改进的Gabor,DLDA和最小二乘向量机进行融合了一种新的算法。在该算法中,首先通过Gabor中引入支持向量的字典学习算法,提高人脸信息,其次在DLDA中采用降低同类中距离偏大的样本之间的类内距离,去除了无用的样本,最后通过最小二乘向量机筛选出优良的样本。在仿真实验中与其他人脸识别算法进行对比,取得了比较好的识别效果。
For the shortcoming of the face image recognition’s low accuracy,the improved Gabor,DLDA and the least square vector machine are integrated in this paper to form a new algorithm.In this algorithm,first,through the introduction of support vector dictionary learning algorithm Gabor,improve the face information,followed by reduction in the large distance between the similar samples within class distance in DLDA,remove the useless samples,and finally,proper samples are selected with the least squares support vector machine.In the simulation experiment,this algorithm is compared with other face recognition algorithms,and good results are achieved.
作者
陈幼芬
Chen Youfen(Department of Electronic and Information Engineering,Shunde Polytechnic,Foshan Guangdong 528333,China)
出处
《科技通报》
2019年第3期113-118,共6页
Bulletin of Science and Technology