基于小波和DFB-PCA的人脸识别算法研究

本文提出了一种基于小波和DFB-PCA 的人脸识别算法。将识别图像先用小波变换对其分解后，再对低频子带进行DFB-PCA 识别。实验结果表明，该方法有利于减少光照、姿势和表情等因素的影响，降低噪声和图像识别的维数，且具有较好的鲁棒性。与直接在原图像上采用PCA 识别相比，可提高对图像的计算速度和识别效率。
关键词：小波变换；DFB-PCA；图像识别；人脸识别

Face Recognition Algorithm Researching Based on Wavelet and DFB-PCA Zhang Jian-ge, Wang Shi-qing, Sheng Guang-lei (College of Information Engineer, Zhengzhou University, Zhengzhou 450002, China) Abstract: In this paper, face recognition algorithm based on wavelet and DFB-PCA which first decomposes recognition images using wavelet transformation and then recognizes the low frequency subband using DFB-PCA algorithm has been proposed. Experimental results demonstrate that the proposed algorithm is propitious to reduce the influence of illumination , pose, expression and so on, reduce the noise of images and the dimension of recognition images, and has a robust performance. The proposed algorithm can improve computing speed of images and the recognition rate compare with the algorithm which uses PCA only on original images.
Key words: wavelet transformation; DFB-PCA; image recognition; face recognition