摘要
目的针对传统PCA在人脸识别过程中对光照敏感、特征提取中出现的耗时长、内存高、首次命中识别率不高等问题进行研究,提出改进方案。方法通过直方图均衡化来减少图像亮度变化带来的影响。使用1种快速PCA方法来加速计算降维过程中的样本协方差矩阵的本征值和本征向量,进一步提出根据特征向量重构测试样本特征向量的方法来提高首次命中识别率。结果将其应用到FERET人脸库进行特征提取,采用多种不同的距离测度进行分类。实验结果验证了该方法在有效降低运算时间的同时能获取较高的识别率。结论提出了1种改进的快速PCA算法进行人脸识别。与传统PCA算法相比,提出的方法减少了计算时间,提高了识别率。
Objective To study the problems of traditional PCA in the process of face recognition,such as long time-consuming,high memory,and low first-time hit recognition rate,and propose improved solutions.Methods The histogram equalization was used to reduce the influence of image brightness variation.A fast PCA method was used to accelerate the calculation of the eigenvalues and eigenvectors of the sample covariance matrix in the dimensionality reduction process.The feature vector of the test sample was reconstructed according to the feature vector to improve the first hit recognition rate.Results The FERET face database was applied to feature extraction and classified by various distance measures.The experimental results showed that the method could effectively reduce the computation time and obtain a higher recognition rate.Conclusion This paper proposes an improved fast PCA algorithm for face recognition.Compared with the traditional PCA algorithm,the proposed method reduces the calculation time and improves the recognition rate.
作者
陈平
CHEN Ping(Department of Electronic Information,Huishang Vocational College,Hefei,Anhui 230000,China)
出处
《河北北方学院学报（自然科学版）》
2020年第5期5-8,共4页
Journal of Hebei North University:Natural Science Edition
基金
2019安徽省教育厅自然科学研究重点项目:“基于卷积神经网络的深度学习算法与图像识别应用研究”(KJ2019A1242)。
