摘要
机器学习中存在大量处理图片的高维数据,PCA是一种有效降维数据的方法.针对PCA算法在提取前几个特征值时计算量大且易受光照噪声等影响的问题,提出一种改进算法,利用分割矩阵的做法求出每一个小矩阵的最大特征值,将其特征向量组成图片的特征矩阵.这样提取出来的特征值更加具有代表性,经仿真实验验证,改进算法的正确识别训练图像集和测试集数目以及识别率均比应用在传统PCA算法上有效.
In machine learning,there are a lot of high-dimensional data processing images and PCA is an effective method to reduce the dimension data.In order to solve the problem that PCA algorithm needs a lot of computation and is easily affected by illumination and noise when extracting the first few eigenvalues,an improved algorithm is proposed.The maximum eigenvalue of each small matrix is obtained by using the method of segmentation matrix,and the corresponding eigenvectors constitute the feature matrix of the image.In this way,the extracted eigenvalues are more representative.The simulation results show that the correct recognition of the number of training image sets and test sets and the recognition rate of the improved algorithm are better than those applied to the traditional PCA algorithm.
作者
岳也
王川龙
YUE Ye;WANG Chuanlong(Department of Mathematics,Taiyuan Normal University,Jinzhong 030619,China)
出处
《太原师范学院学报（自然科学版）》
2021年第1期49-54,68,共7页
Journal of Taiyuan Normal University:Natural Science Edition
关键词
人脸识别
传统PCA算法
特征值提取
BP神经网络算法
face recognition
traditional PCA algorithm
eigenvalue extraction
BP neural network algorithm
