L1-(2D)2PCANet: A Deep Learning Network for Face Recognition

TL;DR

This paper introduces L1-(2D)2PCANet, a deep learning face recognition network utilizing L1-norm-based 2D PCA for filter learning, demonstrating superior robustness to outliers and data variations compared to existing methods.

Contribution

The novel L1-(2D)2PCANet integrates L1-norm 2D PCA into deep learning for face recognition, enhancing robustness to outliers and data changes.

Findings

Outperforms baseline networks on benchmark datasets.

More robust to outliers and image variations.

Achieves higher recognition accuracy.

Abstract

In this paper, we propose a novel deep learning network L1-(2D)2PCANet for face recognition, which is based on L1-norm-based two-directional two-dimensional principal component analysis (L1-(2D)2PCA). In our network, the role of L1-(2D)2PCA is to learn the filters of multiple convolution layers. After the convolution layers, we deploy binary hashing and block-wise histogram for pooling. We test our network on some benchmark facial datasets YALE, AR, Extended Yale B, LFW-a and FERET with CNN, PCANet, 2DPCANet and L1-PCANet as comparison. The results show that the recognition performance of L1-(2D)2PCANet in all tests is better than baseline networks, especially when there are outliers in the test data. Owing to the L1-norm, L1-2D2PCANet is robust to outliers and changes of the training images.