计算机应用 ›› 2010, Vol. 30 ›› Issue (11): 2970-2973.
• 模式识别 • 上一篇 下一篇
火元莲
收稿日期:
修回日期:
发布日期:
出版日期:
通讯作者:
Received:
Revised:
Online:
Published:
摘要： 为了降低光照变化对基于子空间的人脸识别方法性能的影响,结合多通道Log-Gabor策略与(2D)2PCALDA特征提取方法,提出了一种新的人脸识别方法。将不同尺度与方向作为独立通道,在每个通道内采用(2D)2PCALDA对人脸图像的Log-Gabor表示进行特征提取、分类,然后对各通道分类结果进行决策融合得到最终的类别归属。在CAS-PEAL-R1、ORL与Yale 人脸数据库上的实验结果表明,该算法具有较好的识别性能。
关键词: 多通道Log-Gabor, (2D)2PCALDA, 人脸识别
Abstract: To reduce the influence resulting from lighting variations on the performance of face recognition methods based on images, a new face recognition method, which combined the multi-channel strategy with (2D)2PCALDA feature extraction technique, was proposed. Different scale and orientation as an independent channel, the features were extracted using (2D)2PCALDA and classification was performed in each channel, and the final classification could be obtained by fusing different channel classification results. The experimental results in the CAS-PEAL-R1, ORL and Yale face database show that the proposed method has better recognition performance.
Key words: Key words: Multi-channel Log-Gabor, (2D)2PCALDA, Face recognition
火元莲. 基于多通道Log-Gabor小波与(2D)^2PCALDA的人脸识别方法[J]. 计算机应用, 2010, 30(11): 2970-2973.
0 / 推荐
导出引用管理器 EndNote|Ris|BibTeX
链接本文: http://www.joca.cn/CN/
http://www.joca.cn/CN/Y2010/V30/I11/2970
