摘要
姿态变化和光照干扰对于人脸识别的准确率和效率有很大影响。针对这一问题,文中采用结合Gabor特征和SIFT特征的人脸识别方法进行识别,提取一幅人脸图像的多个方向和多个尺度的Gabor特征,并将提取得到的Gabor特征图像进行分块。对分块后的子图像进行提取SIFT特征的操作,将得到的Gabor特征全部SIFT向量级联作为最终特征向量。使用主成分分析方法对得到的最终特征向量进行降维处理,随后使用最小二乘支持向量机进行训练识别。在FERET人脸数据库中进行的实验结果表明,相对于传统单一的人脸识别方法,利用本文方法在姿态变化和光照干扰情况下对人脸识别的准确率达到98.1%,证明了新算法的有效性。
Position variation and illumination interference affect the accuracy and efficiency of face recognition.To solve the problem,a new method of feature extraction combining the Gabor feature and SIFT feature was proposed in this paper.Gabor feature of face images from multiple scales and directions were extracted and divided into sub-images of the same size.The SIFT features were extracted from the partitioned sub-images,and all SIFT vectors of the obtained Gabor features were cascaded as the final feature vectors.Principal component analysis method was used to reduce the dimension of the final eigenvectors,and then the least squares support vector machine was used for training and recognition of images.Experimental results of tests in the FERET face database showed that compared with the traditional single face recognition method,the accuracy of face recognition was 98.1%under the change of position and illumination interference,which proved the effectiveness of the new algorithm.
作者
周柱
甘屹
孙福佳
ZHOU Zhu;GAN Yi;SUN Fujia(School of Mechanical Engineering,University of Shanghai for Science and Technology,Shanghai 200093,China)
出处
《电子科技》
2019年第4期1-5,共5页
Electronic Science and Technology
基金
国家自然科学基金(51375314)~~
