一种基于PCA降维的SIFT算法  

A Dimension Reduction SIFT Algorithm Based on PCA

作　　者：肖璞[1] 殷一鸣 XIAO Pu;YIN Yi-ming(College of Computer Science and Technology,Sanjiang University,Nanjing 210012)

机构地区：[1]三江学院计算机科学与工程学院,南京210012

出　　处：《现代计算机》2020年第34期15-18,43,共5页Modern Computer

基　　金：江苏省高校自然科学研究面上项目(No.17KJD520007)。

摘　　要：人脸识别技术基于人的面部特征提取脸部的几何和颜色等特征以区分个体生物。然而,在人脸图像采集过程中往往受光照强度、噪声强度、旋转、比例缩放、平移、仿射/投影变换、杂物、遮挡和其他环境因素的影响,SIFT算法可解决这一系列问题。但SIFT算法由于提取特征的高维度具有一些缺点,计算复杂度高,计算过程繁琐,影响人脸识别算法的效率。在SIFT特征提取算法的基础上进行改进优化,提出一种基于降维的PCA-SIFT特征提取算法。对面部图像进行收集、分类和预处理,完成并分析多项试验结果对比。最终降低SIFT特征提取算法的维数,提高人脸识别的效率。The face recognition technology is based on human facial features and used to extract features such as face geometry and color to distinguish individual creatures.However,in the process of face image acquisition,it is often affected by illumination intensity,noise intensity,rotation,scaling,translation,affine/projection transformation,debris,occlusion and other environmental factors.SIFT algorithm can solve this series of problems.The SIFT algorithm has some shortcomings due to the high dimension of the extracted features,high computational complexity,complicated calculation process,and affects the efficiency of the face recognition algorithm.This paper optimizes and improves on the basis of SIFT feature extraction algorithm,and proposes a PCA-SIFT feature extraction based on dimension reduction algorithm.At the same time,facial images were collected,classified and preprocessed.The comparison of multiple test results is completed and analyzed.Finally,the dimension of SIFT feature extraction algorithm is reduced,the efficiency of face recognition is improved.

关 键 词：人脸识别 SIFT算法 降维 PCA 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]