基于局部线性嵌入极限学习机的人脸识别新方法

针对人脸图片数量多、容易受噪声干扰,致使人脸识别的识别速度慢、准确率低的问题,提出一种基于局部线性嵌入极限学习机的人脸识别方法——LLE-ELM算法。利用局部线性嵌入(LLE)算法对人脸数据提取特征,最大限度保留原数据的特征结构,减少数据量,降低计算复杂;采用极限学习机(ELM)算法对提取特征后的数据进行分类;实现人脸识别,输出识别准确率和时长。通过在ORL数据库、Yale数据库、AR人脸库和CASIA-WEBFACE人脸库上的数值实验表明:与PCA、SVM、CNN算法对比,该算法具有较高的识别准确率和较快的识别速度。

A NEW FACE RECOGNITION METHOD BASED ON LOCALLY LINEAR EMBEDDING EXTREME LEARNING MACHINE

Due to the large number of face images,which are easy to be interfered by noise,the recognition speed of face recognition is slow and the accuracy is low.We propose a face recognition method based on local linear embedded limit learning machine:LLE-ELM algorithm.The local linear embedding(LLE)algorithm was used to extract features from face data,so as to retain the feature structure of the original data to the maximum extent,decrease the amount of data and reduce the complexity of calculation.The extreme learning machine(ELM)algorithm was used to classify the extracted data.The face recognition,output recognition accuracy and duration were realized.Numerical experiments on the ORL database,Yale database,AR face database and CASIA-WEBFACE face database show that compared with PCA,SVM and CNN algorithms,our algorithm has higher recognition accuracy and faster recognition speed.