A survey on techniques to handle face recognition challenges: occlusion, single sample per subject and expression

Lahasan, Badr; Lutfi, Syaheerah Lebai; San-Segundo, Rubén

doi:10.1007/s10462-017-9578-y

Cited by 60 publications

(22 citation statements)

References 94 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…LBP has been recognised for simple and efficient method for invariant to monotonic greyscale transformation and illumination changes [10]. FARO (Face Recognition against Occlusions and Expression Variations) [11], is an image fractal based technique used for Partitioned Iterated Function System (PIFS), and FACE (Face Analysis for Commercial Entities) [11], which exploit in measuring the local features from the correlation index. Both methods use a localized regionbased like approach.…”

Section: Related Studymentioning

confidence: 99%

See 1 more Smart Citation

Proposed A New Multidimensional Face Recognition For Surgically Altered Faces In Security Surveillance

Chawla¹,

Chandra²

2019

IJEAT

View full text Add to dashboard Cite

From previously carried out review results, authors have proposed a new multidimensional face recognition approach and applied on the sample database of sensitive medical images. The basic idea is to differentiate an individual’s identity with its own core features extracted from local or global plastic surgery. Features are being highlighted to match pre-topost or post-to-pre surgery face images. The procedure may affect different face recognition algorithms working on linear or nonliner variations approach and that’s what multidimensional approach is meant. Hence it proves that the performance of verified rank accuracy ratio has increased in face recognition for security surveillance.

show abstract

Section: Related Studymentioning

confidence: 99%

“…Both methods use a localized regionbased like approach. The FARO performs recognition by dividing face into regions for significant meaning (eyes, nose and mouth) [11]. The second method, it performs the image into different extracted blocks of equal domain FAST Algorithm [12].…”

Section: Related Studymentioning

confidence: 99%

Proposed A New Multidimensional Face Recognition For Surgically Altered Faces In Security Surveillance

Chawla¹,

Chandra²

2019

IJEAT

View full text Add to dashboard Cite

show abstract

“…Recently, Face Recognition (FR) has seen a breakthrough mainly thanks to the introduction of deep neural networks [1,2], thus allowing its adoption in plentiful applications [3]. Even though, there are still several open problems [4] deserving further investigation. The main challenges concern the long-standing difficulties of dealing with images acquired in unconstrained conditions [5], implying the necessity to deal with several illumination conditions, head poses, facial expressions, possible partial occlusions, and possible low image quality [6].…”

Section: Introductionmentioning

confidence: 99%

“…The main challenges concern the long-standing difficulties of dealing with images acquired in unconstrained conditions [5], implying the necessity to deal with several illumination conditions, head poses, facial expressions, possible partial occlusions, and possible low image quality [6]. Furthermore, the matter gets more difficult by the double hardness of accomplishing the recognition task dealing with large-scale databases [7], and having only a few images per subject available for the gallery/train construction, facing the so called Small Sample Size (SSS) problem, or even the extreme case when only one image is available: the Single Sample Per Person (SSPP) problem [4,8]. Such challenge is of leading interest in application such as e-passport control, law enforcement, surveillance, human-computer interaction, to name just a few.…”

Section: Introductionmentioning

confidence: 99%

Robust Single-Sample Face Recognition by Sparsity-Driven Sub-Dictionary Learning Using Deep Features

Cuculo

D’Amelio

Grossi

et al. 2019

Sensors

View full text Add to dashboard Cite

Face recognition using a single reference image per subject is challenging, above all when referring to a large gallery of subjects. Furthermore, the problem hardness seriously increases when the images are acquired in unconstrained conditions. In this paper we address the challenging Single Sample Per Person (SSPP) problem considering large datasets of images acquired in the wild, thus possibly featuring illumination, pose, face expression, partial occlusions, and low-resolution hurdles. The proposed technique alternates a sparse dictionary learning technique based on the method of optimal direction and the iterative sans-serifℓ0-norm minimization algorithm called k-LiMapS. It works on robust deep-learned features, provided that the image variability is extended by standard augmentation techniques. Experiments show the effectiveness of our method against the hardness introduced above: first, we report extensive experiments on the unconstrained LFW dataset when referring to large galleries up to 1680 subjects; second, we present experiments on very low-resolution test images up to 8×8 pixels; third, tests on the AR dataset are analyzed against specific disguises such as partial occlusions, facial expressions, and illumination problems. In all the three scenarios our method outperforms the state-of-the-art approaches adopting similar configurations.

show abstract

“…With the broad establishment in recent years of video surveillance systems and the billions of cameras embedded in smartphones, face analysis from images is an increasingly prevalent task for government agencies and industry alike. While face analysis has been an active research area for several decades, most of the prior work was focused on face verification/identification in relatively constrained environments (e.g., near-frontal poses and under controlled lighting conditions) [1,2].…”

Section: Introductionmentioning

confidence: 99%

Effective and Generalizable Graph-Based Clustering for Faces in the Wild

Chang

Pérez-Suárez

González-Mendoza

2019

Computational Intelligence and Neuroscience

View full text Add to dashboard Cite

Face clustering is the task of grouping unlabeled face images according to individual identities. Several applications require this type of clustering, for instance, social media, law enforcement, and surveillance applications. In this paper, we propose an effective graph-based method for clustering faces in the wild. The proposed algorithm does not require prior knowledge of the data. This fact increases the pertinence of the proposed method near to market solutions. The experiments conducted on four well-known datasets showed that our proposal achieves state-of-the-art results, regarding the clustering performance, also showing stability for different values of the input parameter. Moreover, in these experiments, it is shown that our proposal discovers a number of identities closer to the real number existing in the data.

show abstract

A survey on techniques to handle face recognition challenges: occlusion, single sample per subject and expression

Cited by 60 publications

References 94 publications

Proposed A New Multidimensional Face Recognition For Surgically Altered Faces In Security Surveillance

Proposed A New Multidimensional Face Recognition For Surgically Altered Faces In Security Surveillance

Robust Single-Sample Face Recognition by Sparsity-Driven Sub-Dictionary Learning Using Deep Features

Effective and Generalizable Graph-Based Clustering for Faces in the Wild

Contact Info

Product

Resources

About