关键词: 构造虚拟样本, 教室点名系统, 改进主成分分析-线性判别分析（PCA-LDA）算法, 人脸识别

Abstract: Aiming at the problems in the actual collection of video such as the complex background, changeable characters, long image processing time and the lack of training samples, the improved face recognition algorithm is proposed based on constructing virtual samples and the combination of Gabor filter and PCA-LDA algorithm, in order to be applied to the classroom attendance system. First of all, the video collected in the classroom is cut, and part of the images are captured by the frame and saved as test images. Then it compares the collected test images with the training images in the existing face database, finally usies the proposed algorithm which uses the mirror method to construct virtual samples and combines with improved PCA-LDA algorithm based on Gabor filter and PCNN gray image enhancement algorithm for face recognition. Simulation results show that the proposed algorithm predicts the possible changes in samples, and to some extent, the computational complexity is reduced, the recognition rate is improved obviously, and it is well verified in the classroom attendance system.

Key words: constructing virtual samples, classroom attendance system, improved Principal Component Analysis-Linear Discriminant Analysis（PCA_LDA） algorithm, face recognition