ℓ 2,p-norm based PCA for image recognition

Qianqian Wang, Quanxue Gao, Xinbo Gao, Feiping Nie

Research output: Contribution to journalArticlepeer-review

109 Scopus citations

Abstract

Recently, many ℓ 1-norm-based PCA approaches have been developed to improve the robustness of PCA. However, most existing approaches solve the optimal projection matrix by maximizing ℓ 1-norm-based variance and do not best minimize the reconstruction error, which is the true goal of PCA. Moreover, they do not have rotational invariance. To handle these problems, we propose a generalized robust metric learning for PCA, namely, ℓ 2, p-PCA, which employs ℓ 2, p-norm as the distance metric for reconstruction error. The proposed method not only is robust to outliers but also retains PCA's desirable properties. For example, the solutions are the principal eigenvectors of a robust covariance matrix and the low-dimensional representation have rotational invariance. These properties are not shared by ℓ 1-norm-based PCA methods. A new iteration algorithm is presented to solve ℓ 2, p-PCA efficiently. Experimental results illustrate that the proposed method is more effective and robust than PCA, PCA-L1 greedy, PCA-L1 nongreedy, and HQ-PCA.

Original languageEnglish
Pages (from-to)1336-1346
Number of pages11
JournalIEEE Transactions on Image Processing
Volume27
Issue number3
DOIs
StatePublished - Mar 2018

Keywords

  • Dimensionality reduction
  • Principal component analysis
  • ℓ 1-norm

Fingerprint

Dive into the research topics of 'ℓ 2,p-norm based PCA for image recognition'. Together they form a unique fingerprint.

Cite this