Learning Super-Resolution Coherent Facial Features Using Nonlinear Multiset PLS for Low-Resolution Face Recognition

Face hallucination (FH) is an effective technique for super-resolving low-resolution (LR) face images. In real-world applications, a face image usually has multiple distinct low resolutions. Most existing FH methods can not effectively deal with multiple LR views simultaneously. To solve this issue, we present a multi-set partial least squares (MPLS) approach and its kernel extension for jointly learning the nonlinear consistency of multi-resolution facial features. With nonlinear MPLS, we present a novel simultaneous super-resolution coherent facial feature method for the face images with multiple LRs, which has capacity of jointly learning the nonlinear relationships between multiple facial resolutions. Experimental results demonstrate the effectiveness and robustness of our proposed FH method.

1. INTRODUCTION

Nowadays, face recognition has become a common authentication method in various scientific applications. However, there is an important issue that face recognition system must face in practice. That is, the captured faces are usually of low resolution (LR) because of the influence of distance, lighting, poses, weather, and the limitation of cameras. This leads to the poor performance of traditional face recognition algorithms. To solve this issue, many face hallucination (FH, also called face super-resolution) techniques, have been proposed over the past decade; see, for example, [1], [2], [3], [4].