一种DCT和ELBP融合的人脸特征提取方法
发布时间：2019-06-21 02:25
【摘要】：仅使用单一算法提取人脸图像的特征不足以捕捉人脸多方面的信息,为了更好地获取人脸面部特征,针对离散余弦变换(Discrete Cosine Transform,DCT)只能提取人脸面部图像的频域特征,而未考虑近邻像素之间的关系、不能提取纹理特质信息等问题进行了研究,提出一种融合DCT特征和伸长的局部二值模式(Elongated Local Binary Pattern,ELBP)的特征提取方法。该方法首先考虑将人脸图像经DCT变换后的少量低频系数作为人脸的频域特征,然后对人脸图像中贡献相对较大的眼部和嘴部区域进行ELBP特征提取,将该ELBP特征作为人脸的空域特征,并采用PCA方法对所提取的空频域特征进行有效融合,得到更有效的人脸特征,最后用最近邻分类器进行识别。在ORL人脸库和Yale人脸库上的实验结果表明:所提方法比单独采用DCT、ELBP方法或采用DCT和LBP相结合的方法提取的特征更有利于识别,提高了识别的准确性。
[Abstract]:Using a single algorithm to extract the features of face images is not enough to capture the information of face in many aspects. in order to better obtain face features, discrete cosine transform (Discrete Cosine Transform,DCT) can only extract the frequency domain features of face images, but can not extract texture feature information without considering the relationship between nearest neighbor pixels. A feature extraction method based on local binary pattern (Elongated Local Binary Pattern,ELBP) combining DCT features and extension is proposed. In this method, a small number of low frequency coefficients of face image transformed by DCT are considered as the frequency domain features of face, then the ELBP feature is extracted from the eye and mouth region, and the ELBP feature is regarded as the spatial feature of face. PCA method is used to effectively fuse the extracted space-frequency domain features to obtain more effective face features. Finally, the nearest neighbor classifiers are used for recognition. The experimental results on ORL face database and Yale face database show that the proposed method is more beneficial to recognition and improves the accuracy of recognition than the method of DCT,ELBP alone or the combination of DCT and LBP.
【作者单位】： 兰州理工大学计算机与通信学院;
【基金】：国家自然科学基金(No.61263019) 甘肃省高等学校基本科研项目
【分类号】：TP391.41
[Abstract]:Using a single algorithm to extract the features of face images is not enough to capture the information of face in many aspects. in order to better obtain face features, discrete cosine transform (Discrete Cosine Transform,DCT) can only extract the frequency domain features of face images, but can not extract texture feature information without considering the relationship between nearest neighbor pixels. A feature extraction method based on local binary pattern (Elongated Local Binary Pattern,ELBP) combining DCT features and extension is proposed. In this method, a small number of low frequency coefficients of face image transformed by DCT are considered as the frequency domain features of face, then the ELBP feature is extracted from the eye and mouth region, and the ELBP feature is regarded as the spatial feature of face. PCA method is used to effectively fuse the extracted space-frequency domain features to obtain more effective face features. Finally, the nearest neighbor classifiers are used for recognition. The experimental results on ORL face database and Yale face database show that the proposed method is more beneficial to recognition and improves the accuracy of recognition than the method of DCT,ELBP alone or the combination of DCT and LBP.
【作者单位】： 兰州理工大学计算机与通信学院;
【基金】：国家自然科学基金(No.61263019) 甘肃省高等学校基本科研项目
【分类号】：TP391.41
【相似文献】
相关期刊论文 前10条
1 江宇闻;朱思铭;;基于Overcomplete ICA的人脸特征提取[J];计算机科学;2005年07期
2 陈鹏;;人脸特征提取方法的研究[J];中国水运(理论版);2006年12期
3 魏冬冬;谌海新;聂铁铸;;人脸特征提取与识别技术研究[J];计算机与现代化;2007年03期
4 王宏勇;王青青;;三维人脸特征提取方法综述[J];电子科技;2012年12期
5 王fE;杨光;;改进的脉冲耦合神经网络人脸特征提取方法[J];计算机工程与应用;2013年01期
6 宋宇,周激流,何坤,刘智明,黎奎;多尺度人脸特征提取[J];计算机应用研究;2005年04期
7 陶宇权;程德福;李成荣;徐U，
本文编号：2503712
本文链接：https://www.wllwen.com/kejilunwen/ruanjiangongchenglunwen/2503712.html
