基于改进的加权分块2D-PCA人脸识别技术的研究

Research of the technology of face recognition based on the improved weighted block 2D-PCA

比较了PCA(Principal Component Analysis)和2D-PCA(Two-Dimensional Principal Component Analysis)人脸识别算法。在2D-PCA的基础上提出了一种改进算法,即基于整体区域、感兴趣区域与非感兴趣区域的加权分块2D-PCA算法。该算法借助权值的动态调整,最终实现了最优解。基于知名脸库ORL设计实验来验证文中提出的改进的加权分块2D-PCA算法。分析试验结果表明,发现本算法识别率达到97.5%,较PCA算法提高21.66%,较2D-PCA算法提高10.08%,进一步证实本算法较PCA和2D-PCA显著提高了人脸识别的准确率。

This paper compares the PCA （Principal Component Analysis） and 2D-PCA （Two-Dimen- sional Principal Component Analysis） face recognition algorithm. On the basis of the 2D-PCA, an im- proved weighted block 2D-PCA algorithm was proposed, which is based on the whole region, the re- gion of interest and other regions. By dynamically adjusting the weights, this algorithm finally achieved the optimal solution. To verify the improved weighted block 2D-PCA algorithm proposed in this paper, experiments based on the well-known face database ORL were designed, and the experimental results show that the proposed algorithm recognition rate reached 97.5% , 21.66% higher than that of PCA algorithm, 10.08% higher than that of 2D-PCA algorithm, which further confirmed that compared with PCA and 2D-PCA, the proposed algorithm significantly improves the accuracy of face recognition.