Sparse Illumination Learning and Transfer for Single-Sample Face Recognition with Image Corruption and Misalignment

Abstract

Single-sample face recognition is one of the most challenging problems in face recognition. We propose a novel algorithm to address this problem based on a sparse representation based classification (SRC) framework. The new algorithm is robust to image misalignment and pixel corruption, and is able to reduce required gallery images to one sample per class. To compensate for the missing illumination information traditionally provided by multiple gallery images, a sparse illumination learning and transfer (SILT) technique is introduced. The illumination in SILT is learned by fitting illumination examples of auxiliary face images from one or more additional subjects with a sparsely-used illumination dictionary. By enforcing a sparse representation of the query image in the illumination dictionary, the SILT can effectively In particular, the single-sample face alignment accuracy is comparable to that of the well-known Deformable SRC algorithm using multiple gallery images per class. Furthermore, the face recognition accuracy exceeds those of the SRC and Extended SRC algorithms using hand labeled alignment initialization.

A Customized Sparse Representation Model With Mixed Norm for Undersampled Face Recognition

“…We compare the proposed method with several related methods: SRC [16], CRC [20], LRC [15], AGL [26], ESRC [27], SVDL [30], SILT [31] and LRAGL [29] on image pixel and compare some handcrafted feature method (Gabor [48], local binary pattern (LBP) [49], [50] and discriminant face descriptor (DFD) [51]). In order to show the performance Fig.7(b).…”

“…It has been proved that ESRC improves efficiently the performance for undersampled FR with non-idea conditions. In addition, there are other typical methods, such as SVDL [30] and SILT [31], which add the learned sparse dictionary into the ESRC framework.…”

“…SVDL requires a large enough generic set, in which each subject should contain multiple face images with all types of variations. In 2014, Zhuang et al [31] proposed a new dictionary learning method, i.e. sparse illumination learning and transfer (SILT), to learn from the generic set independently.…”

Semi-Supervised Sparse Representation Based Classification for Face Recognition With Insufficient Labeled Samples

“…Motivated by ESRC, Yang et al proposed the Sparse Variation Dictionary Learning (SVDL) model to learn the variation dictionary V, more precisely [14]. In addition to modeling the variation dictionary by a linear illumination model, Zhuang et al [15], [16] also integrated auto-alignment into their method. Gao et al [26] extended the ESRC model by dividing the image samples into several patches for recognition.…”

Robust Face Recognition Based on DCNN and CRC

“…This method has a little improvement in the misalignment circumstance, but it failed to obtain a good recognition result in the pose variation at 25 degrees. More recently, Liansheng Zhuang et al [6] has introduced a sparse illumination learning and transfer (SILT) technique into SRC to address the situation of single sample regime and fewer restrictions. But it has just considered the circumstance of rotation and scale, failed to take pose variation into account.…”