基于LPP_SIFT和巴氏距离相结合的人脸识别算法

摘要

局部保持映射_尺度不变特征变换（LPP-SIFT）算法是一种有效的特征识别方法。提出了奇异值分解的LPP-SIFT和巴氏距离相合的算法。基于LPP的算法在人脸识别中容易遇到奇异值问题。为此，采用奇异值分解的LPP-SIFT算法进行特征提取和降维处理，然后采用巴氏距离特征的迭代算法，得到最小错误率上界。在ORL上实验，实验结果验证了提出算法在人脸识别中的有效性。%Locality Preserving Projection-SIFT is an effective method which can extract the feature.This paper proposes a method which combines LPP-SIFT of Singular Value Decomposition (SVD) and Bhattacharyya distance. based LPP is known to suffer from singular value problem. Therefore, using LPP-SIFT of SVD was used to extract feature and reduce dimension. Then it gets the smallest error rate upper bound by using iterative algorithm which has Bhattacharyya distance feature. The experimental results demonstrate the efficacy of the proposed approach for face recognition on ORL.