摘要
针对人脸识别中的特征提取问题,提出一种基于Log-Gabor和正交等度规映射(Orthogonal IsoProjection,OIsop)的人脸识别算法。算法首先采用Log-Gabor小波对图像进行滤波来提取高阶非线性统计信息。然后,在原始的优化问题中增加正交约束条件,推导出能得到一组具有正交性最优映射向量的迭代公式,使得算法更利于保留人脸非线性子流形空间与距离有关的结构信息和重构样本。通过ORL和PIE库上的人脸识别实验验证了算法的有效性。
In view of the problems of feature extraction in face recognition,a Log-Gabor and orthogonal supervised IsoProjection based algorithm for face recognition was proposed in this paper.The proposed algorithm first gets the high-order nonlinear statistical information by calculating the Log-Gabor wavelet representation of face images.Then the orthogonal constrained conditions added to the original optimal problem and the iterative formulae for finding a set of orthogonal optimal projection vectors are deduced.The orthogonal basis can help to preserve the information of nonlinear sub-manifold space which is related to distance and reconstruct data.The experimental results on ORL and PIE face database illustrate the effectiveness of the proposed method.
出处
《计算机科学》
CSCD
北大核心
2011年第2期274-276,295,共4页
Computer Science
基金
国家高技术研究发展计划(863)项目(2009AA04Z215)资助。
