l 2,p -norm sequential bilateral 2DPCA: a novel robust technology for underwater image classification and representation
- PDF / 1,802,391 Bytes
- 15 Pages / 595.276 x 790.866 pts Page_size
- 14 Downloads / 184 Views
(0123456789().,-volV)(0123456789(). ,- volV)
ORIGINAL ARTICLE
l2,p-norm sequential bilateral 2DPCA: a novel robust technology for underwater image classification and representation Pengfei Bi1 • Jian Xu1 • Xue Du1 • Juan Li1 • Guangjia Chen1 Received: 27 November 2019 / Accepted: 7 April 2020 Ó Springer-Verlag London Ltd., part of Springer Nature 2020
Abstract Many recently proposed robust two-dimensional principal component analysis (2DPCA) approaches can suppress the sensitivity to outliers in images to some extent. However, most approaches can neither perfectly minimize the reconstruction error nor use fewer coefficients to conveniently represent image information. To alleviate these deficiencies, we developed a novel robust 2DPCA approach for underwater image analysis, called l2,p-sequential bilateral-2DPCA (l2,p-SB2DPCA). The outstanding advantages of l2,p-SB-2DPCA are as follows. First, our model uses the l2,p-norm as the metric criterion of the objective function, which not only improves the robustness of the algorithm but also preserves the basic properties of 2DPCA. Second, we establish the relationship between the variance of the projection data and the corresponding input data in both the row and column directions, which makes the model achieve good recognition ability while using fewer coefficients and further improves the interpretability of the model. Finally, to obtain the optimal value of l2,pSB-2DPCA, we present an iterative algorithm. The experimental results show that the proposed algorithm achieves the best performance on three underwater datasets and one extended face dataset compared with other robust 2DPCA approaches. Keywords Underwater image analysis l2,p-norm Two-dimensional principal component analysis (2DPCA) Sequential bilateral
1 Introduction In recent decades, the development of underwater optical technology has proceeded extremely rapidly, thereby expanding the research scope of computer vision and pattern recognition from atmospheric images to underwater images [1, 2]. However, most optical images have a common feature called high-dimensional characteristics. & Jian Xu [email protected] Pengfei Bi [email protected] Xue Du [email protected] Juan Li [email protected] Guangjia Chen [email protected] 1
College of Automation, Harbin Engineering University, Harbin 150001, Heilongjiang, China
Therefore, many dimensionality reduction techniques have been developed to solve this problem, among which principal component analysis (PCA) [3, 4] linear discriminant analysis (LDA) [5], locality preserving projection (LPP) [6] and neighborhood preserving projections (NPP) [7] are the most representative approaches. PCA is employed to obtain features that contain the most important information about the data, and LDA can extract the most discriminative features embedded in the data. Both PCA and LDA protect the global geometric information of the data well. Compared with PCA and LDA, LPP and NPP are more focused on revealing the intrinsic geometric information of the data space
Data Loading...
