Disguise detection and face recognition in visible and thermal spectrums
Citations
366 citations
Cites background from "Disguise detection and face recogni..."
...Some initial efforts to study thermal imaging for liveness detection have already been carried out [159], including the acquisition of a significantly large database of thermal images for standard and disguised access attempts where very promising results have been obtained [122]....
[...]
335 citations
139 citations
114 citations
Cites background from "Disguise detection and face recogni..."
...sting and challenging covariate of face recognition. It includes intentional or unintentional changes through which one can either hide his/her identity or appear to be someone else. Dhamecha et al. [Dhamecha et al. 2013] have summarized the existing disguise detection and face recognition algorithms. In this survey, we rather focus on matching across plastic surgery variations. With reduction in cost and time requir...
[...]
111 citations
Additional excerpts
...Currently, feature extraction algorithms such as gray-scale value histogram features [30], thermal local binary pattern (LBP), SIFT features [31], and speeded up robust features (SURF) [32, 33], are widely applied in thermal face recognition....
[...]
...Currently, feature extraction algorithms such as gray-scale value histogram features [30], thermal local binary pattern (LBP), SIFT features [31], and speeded up robust features (SURF) [32], [33], are widely applied in thermal face recognition....
[...]
References
37,861 citations
"Disguise detection and face recogni..." refers methods in this paper
...For every patch, a decision score s is computed using SVM....
[...]
...The normalized descriptor is provided as input to SVM with Radial Basis Function kernel for patch classification....
[...]
...Face recognition with biometric patches is classified using ITE and SVM classifier 2....
[...]
...An SVM model is learned from the ITE descriptors of all the patches from training images (which are annotated manually)....
[...]
...Therefore, in this research, we have proposed the use of Support Vector Machine (SVM) [8], a discriminative classifier, for classifying biometric and non-biometric patches....
[...]
9,658 citations
"Disguise detection and face recogni..." refers methods in this paper
...The proposed approach improves verification accuracy over direct application of LBP, COTS, and SRC....
[...]
...In case of SRC, for fusion of two spectrums, sum rule fusion of residuals of both individual spectrums is performed....
[...]
...Note that, SRC is one of the most important techniques in literature for addressing occlusion/disguise....
[...]
...The result section is divided into three subsections: patch classification using ITE, evaluation of the proposed algorithm, and comparison with sparse representation classifier (SRC) and COTS....
[...]
...As shown in Table 2, although the performance reported by the proposed approach is not as high as it is usually reported in face recognition literature, it outperforms one of the state-of-art commercial system, and a widely used technique, SRC, by at least 3% at 1.0% FAR....
[...]
6,384 citations
[...]
3,650 citations
"Disguise detection and face recogni..." refers background in this paper
...Further, the visible spectrum databases generally used for disguise related research (AR [13] and Yale [2] face databases) contain very limited disguise variations, such as scarves and/or sunglasses....
[...]
...AR database [13] contains 3200+ images pertaining to 126 subjects with two kinds of disguises (sunglasses and scarves)....
[...]
...Note that most of the research has been performed in visible spectrum using the AR [13] and Yale [2] face databases which contains very limited disguise (sunglasses and scarves only)....
[...]
2,952 citations
"Disguise detection and face recogni..." refers background in this paper
...Further, the visible spectrum databases generally used for disguise related research (AR [13] and Yale [2] face databases) contain very limited disguise variations, such as scarves and/or sunglasses....
[...]
...AR database [13] contains 3200+ images pertaining to 126 subjects with two kinds of disguises (sunglasses and scarves)....
[...]
...Note that most of the research has been performed in visible spectrum using the AR [13] and Yale [2] face databases which contains very limited disguise (sunglasses and scarves only)....
[...]