Showing papers on "Periocular Region published in 2019"

Predicting Gender From Iris Texture May Be Harder Than It Seems

Abstract: Predicting gender from iris images has been reported by several researchers as an application of machine learning in biometrics. Recent works on this topic have suggested that the preponderance of the gender cues is located in the periocular region rather than in the iris texture itself. This paper focuses on teasing out whether the information for gender prediction is in the texture of the iris stroma, the periocular region, or both. We present a larger dataset for gender from iris, and evaluate gender prediction accuracy using linear SVM and CNN, comparing hand-crafted and deep features. We use probabilistic occlusion masking to gain insight on the problem. Results suggest the discriminative power of the iris texture for gender is weaker than previously thought, and that the gender-related information is primarily in the periocular region.

Simultaneous Iris and Periocular Region Detection Using Coarse Annotations

Abstract: In this work, we propose to detect the iris and periocular regions simultaneously using coarse annotations and two well-known object detectors: YOLOv2 and Faster R-CNN. We believe coarse annotations can be used in recognition systems based on the iris and periocular regions, given the much smaller engineering effort required to manually annotate the training images. We manually made coarse annotations of the iris and periocular regions (≈122K images from the visible (VIS) spectrum and ≈38K images from the near-infrared (NIR) spectrum). The iris annotations in the NIR databases were generated semi-automatically by first applying an iris segmentation CNN and then performing a manual inspection. These annotations were made for 11 well-known public databases (3 NIR and 8 VIS) designed for the iris-based recognition problem, and are publicly available to the research community1. Experimenting our proposal on these databases, we highlight two results. First, the Faster R-CNN + Feature Pyramid Network (FPN) model reported an Intersection over Union (IoU) higher than YOLOv2 (91.86% vs 85.30%). Second, the detection of the iris and periocular regions being performed simultaneously is as accurate as performed separately, but with a lower computational cost, i.e. two tasks were carried out at the cost of one.

CNN-based gender classification in near-infrared periocular images

Abstract: Periocular region has emerged as a key biometric trait with potential applications in the forensics domain. In this paper, we explore two convolutional neural network (CNN)-based approaches for gender classification using near-infrared images of the periocular region. In the first stage, our approaches automatically detect and extract left and right periocular regions. The first approach utilizes a domain-specific pre-trained CNN to extract deep features from the periocular images. A trained support vector machine (SVM) then utilizes these features to predict the gender information. The second approach employs an end-to-end classifier obtained by fine-tuning a pre-trained CNN on the periocular images. Performance evaluations have been carried out on three databases, which includes an in-house and two public databases. Local binary pattern and histogram of oriented gradient-based methods have been used as baseline methods to ascertain the effectiveness of the proposed approaches. Our results indicate that the proposed approaches achieve higher classification accuracy than the baseline methods, particularly on one of the public databases that contains a large number of non-ideal images. In addition, accuracy of the proposed approaches is consistently higher than the existing eyebrow feature-based method.

Periocular cutaneous sarcoid: case series and review of the literature.

Abstract: Objectives To illustrate the varying clinical presentations of cutaneous sarcoidosis affecting the periocular region, which may masquerade as other clinical entities such as basal cell carcinoma or seborrheic dermatitis. Furthermore, the authors present an unusual observation of lupus pernio involving the adnexal region with the rare presence of perineural granulomas on histology following incisional biopsy. Methods We report a consecutive series of four cases with lesions involving the eyelids with varying clinical appearances. All four patients presented to our adnexal service undergoing incisional diagnostic biopsy. Histology following biopsy subsequently resulted in further investigation and management of both local cutaneous lesions and systemic sarcoidosis. Results Three of our four cases had evidence of pulmonary involvement on chest X-ray. Over an 18-month period, one of two patients responded to intralesional triamcinolone and subsequently to oral methotrexate (15 mg/week). Two patients were observed with their periocular lesions remaining stable without therapy. Conclusions All four patients presented to the adnexal service with lesions of varying morphology and were diagnosed with sarcoidosis following incisional biopsy highlighting the vital role of oculoplastic surgeons in diagnosing this multisystem inflammatory disease. We describe our experience of intralesional triamcinolone, oral methotrexate and watchful observation in the management of such lesions.

Providing Optimal Rejuvenation to the Periocular Area Using Botulinum Toxin A Neuromodulators and Hyaluronic Acid Dermal Fillers

Abstract: The periorbital area is the first area of the face to show signs of aging. To provide safe and natural looking rejuvenation of the delicate eyelids, and supporting structures, an advanced understanding of anatomy, ideal facial proportions, and the most effective methods for rejuvenation is discussed. Periocular rejuvenation is particularly challenging due to the intricate and delicate anatomy of the periocular area. To ensure safe and successful outcomes, it is crucial that injectors use a global approach when providing treatments and that they consider soft tissue, vasculature, and bone structure of the periocular region before administering treatments for aesthetic rejuvenation. Neuromodulators, specifically botulinum toxin A (BoNT-A), and hyaluronic acid (HA) dermal fillers are 2 nonsurgical treatments frequently used to address signs of aging in the periocular area. The objective of this article is to review different BoNT-A and HA filler treatments and discuss how these treatments can be used for optimal rejuvenation of the periocular area.

Relevant features for gender classification in NIR periocular images

Abstract: Most gender classifications methods from near-infrared (NIR) images have used iris information. Recent work has explored the use of the whole periocular iris region which has surprisingly achieved better results. This suggests the most relevant information for gender classification is not located in the iris as expected. In this work, the authors analyse and demonstrate the location of the most relevant features that describe gender in periocular NIR images and evaluate their influence in classification. Experiments show that the periocular region contains more gender information than the iris region. They extracted several features (intensity, texture, and shape) and classified them according to their relevance using the XgBoost algorithm. Support vector machine and nine ensemble classifiers were used for testing gender accuracy when using the most relevant features. The best classification results were obtained when 4000 features located on the periocular region were used (89.22%). Additional experiments with the full periocular iris images versus the iris-occluded images were performed. The gender classification rates obtained were 84.35 and 85.75%, respectively. From results, they suggest focusing only on the surrounding area of the iris. This allows us to realise a faster classification of gender from NIR periocular images.

Extract Features from Periocular Region to Identify the Age Using Machine Learning Algorithms

Abstract: Latest studies done on huge data collected from aging features proved that the performance of facial image based age estimation is low and need to be improved. One of the significant biometric traits for human recognition or search is Human age. Age assessment is very much exigent over other pattern recognition problems since the aging differs from person to person. This paper proposes a new framework that uses periocular region for age feature extraction and application of hybrid algorithm for age recognition. Firstly, preprocessing and periocular region normalization is done to acquire age invariant features. Secondly, the periocular region that underwent preprocessing is analyzed using hybrid approach, a novel machine algorithm that combines both SVM and kNN. The proposed technique generates the best recognition outputs.

Periocular Histiocytoid Carcinoma: Potential Diagnostic Challenges.

Abstract: Cutaneous histiocytoid carcinoma can occur as a primary tumor of the periocular region Morphologically similar histiocytoid carcinomas arising as primary tumors of the breast have a predilection for orbital metastases They can occasionally contain regions with prominent vacuolated cytoplasm and minimal nuclear atypia, which mimic benign histiocytic lesions Differentiating nonneoplastic, primary neoplastic, and metastatic histiocytoid lesions involving the periorbita can be challenging for both the clinician and the pathologist, and this distinction has management implications Herein, we present 3 cases to illustrate the challenges of diagnosing periocular histiocytoid carcinoma

CNN Based Periocular Recognition Using Multispectral Images

Abstract: Over the recent years, the periocular region has emerged as a potential unconstrained biometric trait for person authentication. For a biometric identification scenario to operate reliably round the clock, it should be capable of subject recognition in multiple spectra. However, there is limited research associated with the non-ideal multispectral imaging of the periocular trait. This is critical for real life applications such as surveillance and watch list identification. The existing techniques for multispectral periocular recognition rely on fusion at the feature level. However, these handcrafted features are not primarily data driven and there even exists possibilities for more novel features that could better describe the same. One possible solution to address such issues is to resort to the data driven deep learning strategies. Accordingly, we propose to apply the attributes extracted from pretrained CNN for subject authentication. To the best of our knowledge, this is the first study of multispectral periocular recognition employing deep learning. For our work, the IIITD Multispectral Periocular (IMP) database is used. The best classification accuracy reported for this dataset is 91.8%. This value is not precise enough for biometric identification tasks. The off-the-shelf CNN features employed in our work gives an improved accuracy of 97.14% for the multispectral periocular images.

Periocular Recognition Under Unconstrained Conditions Using CNN-Based Super-Resolution

Abstract: A major challenge of the prevailing biometric systems is the short range of trait capture. Relaxing the range constraints imposed on the subjects arouses advanced challenges in the quality of image and resolution. In such scenarios, the lack of quality of acquired biometric information can be addressed using super-resolution, a technique of generating high-resolution images from low resolution counterparts. Procurement of the periocular images requires less cooperation of subjects compared to other ocular biometrics, thereby emerging as a reliable trait for unconstrained biometrics. For our work, UBIRIS v.2 is used which provides the relevant data for less constrained biometrics. The best rank-1 accuracy reported for periocular recognition with this database is 87.62%, and such works rely on the local and global feature descriptors. On this account, we cannot possibly arrive at the conclusion that these attributes in the literature are the best descriptors for the periocular region. One possible solution to achieve better recognition performance is to employ the latest trends in deep learning. Accordingly, we propose to apply deep learning-based super-resolution technique to the periocular images for improved identification efficiency. Our research work is novel in the prospect that this is the first study for periocular recognition applying CNN-based super-resolution. The off-the-shelf CNN features used in our work give improved rank-1 accuracy of 91.47%.

Relevant features for Gender Classification in NIR Periocular Images

Abstract: Most gender classifications methods from NIR images have used iris information. Recent work has explored the use of the whole periocular iris region which has surprisingly achieve better results. This suggests the most relevant information for gender classification is not located in the iris as expected. In this work, we analyze and demonstrate the location of the most relevant features that describe gender in periocular NIR images and evaluate its influence its classification. Experiments show that the periocular region contains more gender information than the iris region. We extracted several features (intensity, texture, and shape) and classified them according to its relevance using the XgBoost algorithm. Support Vector Machine and nine ensemble classifiers were used for testing gender accuracy when using the most relevant features. The best classification results were obtained when 4,000 features located on the periocular region were used (89.22\%). Additional experiments with the full periocular iris images versus the iris-Occluded images were performed. The gender classification rates obtained were 84.35\% and 85.75\% respectively. We also contribute to the state of the art with a new database (UNAB-Gender). From results, we suggest focussing only on the surrounding area of the iris. This allows us to realize a faster classification of gender from NIR periocular images.

Ocular Recognition Databases and Competitions: A Survey

Abstract: The use of the iris and periocular region as biometric traits has been extensively investigated, mainly due to the singularity of the iris features and the use of the periocular region when the image resolution is not sufficient to extract iris information. In addition to providing information about an individual's identity, features extracted from these traits can also be explored to obtain other information such as the individual's gender, the influence of drug use, the use of contact lenses, spoofing, among others. This work presents a survey of the databases created for ocular recognition, detailing their protocols and how their images were acquired. We also describe and discuss the most popular ocular recognition competitions (contests), highlighting the submitted algorithms that achieved the best results using only iris trait and also fusing iris and periocular region information. Finally, we describe some relevant works applying deep learning techniques to ocular recognition and point out new challenges and future directions. Considering that there are a large number of ocular databases, and each one is usually designed for a specific problem, we believe this survey can provide a broad overview of the challenges in ocular biometrics.

Periocular Region Based Biometric Identification Using SIFT and SURF Key Point Descriptors

Abstract: In the proposed paper the discriminative capability of the periocular region of the face area is exploited in designing a robust biometric system. A periocular region-based biometric system is emerging as a better alternative for the unconstrained and uncontrolled environments. In the projected work the two key point descriptors Speeded Up Robust Features (SURF) and Scale Invariant Feature Transform (SIFT) are employed for deriving the distinctive characteristics from the periocular and complete face portions. City Block Distance method is used to compare the two feature vectors obtained from the SURF and SIFT methods. FERET and FRGC databases are used to estimate the merits of both face and periocular biometric methods. By utilizing only 25% of entire face, periocular area-based biometric systems provided almost the similar performance equivalent to the face biometric systems.

Periocular Lobular Capillary Hemangioma in Adults: A Clinicopathological Study.

Abstract: PURPOSE: To describe a case series of periocular lobular capillary hemangiomas in adults, outlining characteristic clinical and histopathological patterns. MATERIALS AND METHODS: This was a retrospective case series of 16 patients with review of clinical and histopathological features. RESULTS: Eleven male and five female patients were diagnosed with periocular lobular capillary hemangioma at a median age of 38 years (mean, 41 years; range, 21–71 years). The median tumor basal diameter was 6 mm (mean, 7 mm; range, 3–14 mm) and all were well circumscribed. They arose over the course of weeks to months and developed most commonly in the eyelid region (n = 10), followed by the conjunctiva (n = 6). Excisional biopsy of the lesion was done in all cases. On histopathology, the tumors were composed of repeating units of capillary-sized lobules lined by plump endothelial cells. Lesion recurrence was noted in one case. CONCLUSION: Lobular capillary hemangiomas are common benign vascular tumors of periocular region in adults. Clinicohistopathological features and clinical presentation of these lesions are distinctive. Excisional biopsy was curative with recurrence noted rarely.

Novel application of anterior segment optical coherence tomography for periocular imaging.

Abstract: Objective This study aims to assess the use of optical coherence tomography (OCT) imaging for periocular skin lesions and to determine which characteristic features of these images can be correlated to histopathology. Design This is an ongoing prospective study with Research Ethics Board approval. Participants Fifty patients over 18 years old with lesions clinically suspicious of nonmelanoma skin cancer on the periocular region were included in this study. Methods After consent was obtained, clinical photographs and dermatoscopic images were obtained (DermLite II Hybrid M) from the lesion and its contralateral side. Subsequently, the patient was subjected to OCT imaging using the anterior segment module of a spectral domain OCT (Optovue Avanti) and images of the contralateral skin were also obtained. Surgical excision of the lesion was performed and sent for histopathological examination as per routine treatment. OCT images were then correlated to their matching digitalized histopathology section (Philips Ultra Fast Scanner 1.6 RA). Results Based on the OCT images acquired from 50 patients, 8 predominant architectural features have been correlated to histopathology: hyperkeratosis, acanthosis, loss of dermal-epidermal junction delineation, hyporeflective tumour nests, cystic structures, “bunch of grapes” nodules, hyperreflective nests, and ulcerations. Results observed from 45 malignant lesions (basal cell carcinoma, squamous cell carcinoma, and sebaceous gland carcinoma) suggest that groups of features and their layout within the same OCT image may be associated to specific tumour characteristics. Conclusions Current data suggest that anterior-segment OCT imaging is a noninvasive imaging modality for periocular lesions and may be a valuable tool to help differentiate between some tumour types before a biopsy is performed.

Periocular Region-Based Biometric Identification Using Local Binary Pattern and Its Variants

Abstract: Biometric systems are gaining importance significantly in the present day automated systems especially in the areas of authentication, access control, security, and forensic applications for the identification of criminals. Existing biometric systems using the face and iris regions had reached the state of maturity with almost having high performances of 100% accuracy provided the images of the subjects are acquired in the cooperative scenarios. The periocular region is one of the most promising biometric traits providing better robustness and high discrimination ability. Periocular region-based biometric recognition systems are well suited for the wild environments where the subjects are not cooperative. In the proposed paper, the pixel-based LBP and patch-based LBP variants are used as local descriptors for the feature extraction of discriminative features from the full face and periocular regions. Euclidean distance is used to find the matching score between two extracted feature vectors. The experimentation is performed on FRGC, FERET, and Georgia Tech face databases to compare the performance of both periocular and face biometric modalities. It showed that the periocular region has almost the same level of performance of the face region using only 25% data of the complete face.

Orthogonal rotation invariant features for iris and periocular recognition

Abstract: In a non-ideal scenario, iris recognition becomes challenging due to occlusion noise by eyelashes and eyelids, specular reflections and illumination variations. This limits its applicability to be used in real-time applications. Thus, periocular recognition is used in complementary to iris recognition which refers to the region around eyes including eyelashes, eyelids and skin texture. By fusing both iris and periocular modalities, a more reliable and an accurate biometric system is attained that can be considered for high surveillance applications. The proposed techniques are based on continuous orthogonal moments: Zernike moments and polar harmonic transforms which are invariant to rotation and noise. These capture local intensity variations of the neighbourhood pixels that pertain to shape details of the periocular region and random texture pattern of the iris region. The techniques have been evaluated on iris databases: IITD v1 and UBIRIS v2 and a self-developed PEC, Chandigarh periocular database which has been created in a less constrained environment for the research community working on periocular recognition. Results demonstrate that the proposed technique gives encouraging results in comparison to the existing approaches.

Deep Periocular Recognition: A Case Study

Abstract: The periocular region of a face can be used as an autonomous modality in a biometric recognition system. We evaluate two different deep learning pipelines, one with a specific segmentation step and one without it, and show the positive and negative properties of both of them. The obtained results on the newly introduced public dataset SBVPI show that the periocular region offers enough distinguishing information for successful identity recognition.

Impact of Facial Expressions on the Accuracy of a CNN Performing Periocular Recognition

Abstract: The biometric periocular trait refers to the face region in the vicinity of the eye, including the eyelids, eyelashes and eyebrows. The periocular region has emerged as a promising trait for unconstrained biometrics, due to recent advances of convolutional neural networks and the demand for robust face or iris recognition systems. The periocular region can offer global information about the eye shape, and about the texture of the iris, sclera and skin around the eyes. However, periocular biometrics is a relatively new area of research. Thus, it's important to understand the uniqueness and stability of this trait, taking into account the best accuracies obtained by deep learning methods applied on biometric image recognition. In this work, we investigate if changes in the periocular region, caused by facial expressions, affect the recognition accuracy. We apply an existing pretrained CNN architecture, called MobileNet, to the task of periocular recognition. The periocular images used in the experiments were extracted from the Extended Cohn-Kanade expression database. The best results were obtained when the network was tested with similar samples to those contained in the training set. We concluded that the CNN is sensitive to changes in the facial expressions and samples of all expressions are required for training aiming the best accuracy.

Comparative Analyses of Tumour Volume Doubling Times for Periocular and Non-periocular Head and Neck Basal Cell Carcinomas.

Abstract: Basal cell carcinomas are the commonest solid malignancy in humans and thought to grow faster in the periocular region. We measured growth rates between periocular and non-periocular nodular basal cell carcinomas in the head and neck region from high-resolution digital photos and operative notes. The non-periocular basal cell carcinomas (head and neck) showed a mean tumour volume doubling time of 129.8 ± 21.74 (n = 79) days, and the periocular basal cell carcinoma a mean of 177.5 ± 37.21 (n = 47) days. The unpaired t-test with Welch correction showed that this difference was not significant (p = 0.2719). The mean tumour volume doubling time was 147.59 ± 37.75 days for head and neck basal cell carcinomas overall. For the first time, tumour volume doubling times for nodular basal cell carcinomas in the periocular versus non-periocular regions for the head and neck area were analysed, with no significant differences demonstrated. Further, comparison of basal cell carcinoma growth rates with other common solid tumours confirmed that basal cell carcinomas are slow growing malignancies.

Nonsurgical management of periocular scars along with review of literature

Abstract: Introduction: Facial scars in the periocular region may cause cosmetic blemish as well as functional deficits. There are several treatment options available for scar management, but many of these are unsuitable for periocular region because of their potential to cause damage to the ocular structures. This article aims to present an ophthalmic plastic surgeon's perspective of managing these scars safely and effectively. Methods: An extensive literature search was done using PubMed (Medline), Cochrane, and Google Scholar with multiple combinations of search terms such as scar, facial scar, contracture, and keloid. Articles in English language published describing the existing and emerging modalities of treatment for periocular facial scars were reviewed. Results: The scarring pattern in the face and periocular area is different as wound healing in these areas differs from that of the rest of the body. Various techniques ranging from simple scar massage to laser and intralesional steroids and antimetabolites have been described with good results for managing scars in various parts of the body. However, safety of some of these modalities in periocular region has not been established unequivocally as yet. Conclusion: Effective scar management requires a multipronged therapeutic approach. Facial and periocular scars deserve special care due to their close proximity to the eyes. Hence, it is paramount that safety of any approach be ascertained prior to the procedure, especially in periocular areas.

Primary cutaneous pleomorphic rhabdomyosarcoma in the periocular region.

Abstract: Cutaneous pleomorphic rhabdomyosarcoma (PRMS) is a rare malignant mesenchymal tumour which can affect all age ranges, though subcutaneous forms of this disease are uncommonly described. We present a case of a 77-year-old lady with a rapidly growing periocular lesion subsequently demonstrated on histopathological and immunohistochemical examination to be a PRMS. No orbital involvement was demonstrated and no distant metastases were evident. PRMSs represent highly aggressive and infiltrative tumours and must be recognized in a timely fashion to allow safe and adequate treatment. Alongside traditional histopathological examination, immunohistochemistry is invaluable in the diagnosis of this condition. PRMS is a mesenchymal tumour, which classically demonstrates skeletal muscle differentiation. Primary cutaneous PRMS is a rare variant of adult sarcoma, arising most commonly in the head and neck, or the soft tissues of the extremities. They occur almost exclusively in individuals over 45 years of age. PRMS arising in the dermis and subcutis is extremely rare, with few cases having been reported in the literature. To our knowledge, there are no previous case reports of adult primary cutaneous PRMS occurring in the periocular region in the English literature, and only one case is described in the French literature. A 77-year-old female presented with a 50 mm × 40 mm painless lesion of the left sub-brow area that had rapidly enlarged since appearing 3 months prior (Fig. 1). No palpable regional lymph nodes were detected. Magnetic resonance imaging did not demonstrate orbital invasion and a chest X-ray was normal. The lesion was removed in a one-stage excision and reconstruction, with histological examination of the mass revealing a pleomorphic spindle cell formation without evidence of lymphovascular or perineural infiltration. The lesion contained atypical spindle and epithelioid cells, with abundant mitoses. There was no connection to the overlying epidermis. Follow-up at 7 months demonstrated no clinical evidence of local recurrence and no distant disease. Immunohistochemical stains were positive for desmin (Fig. 2) and myogenin, and negative for CD31, H Caldesmon, CD68, S100, SOX10, pan melanin and MNF116. Myogenin stain was subsequently found to be positive. Rhabdomyosarcoma is a rare and aggressive tumour that can be found in any part of the body where skeletal muscle is present, though this is not a prerequisite to tumour formation. Rhabdomyosarcoma can broadly be split into four main histological sub types: pleomorphic, embryonal, alveolar and spindle cell/sclerosing. PRMS is the rarest subtype, occurring most commonly in elderly male patients. PRMS arising in the dermis and subcutis is rarely described. PRMS is most commonly found within the paediatric population, and is readily described within the orbit. Only one previous case of palpebral PRMS in the English literature is described, which occurred in a child. One further case of primary palpebral PRMS is described in the French literature. Similar to our case, both tumours grew rapidly and had a firm texture and blue discolouration. The rapidity of growth may help differentiate primary cutaneous PRMS from other malignant skin lesions. Histopathologically, PRMS is composed of large, atypical spindle cell rhabdomyoblasts, with abundant eosinophilic cytoplasm. Immunostains are

Simultaneous Iris and Periocular Region Detection Using Coarse Annotations

Abstract: In this work, we propose to detect the iris and periocular regions simultaneously using coarse annotations and two well-known object detectors: YOLOv2 and Faster R-CNN. We believe coarse annotations can be used in recognition systems based on the iris and periocular regions, given the much smaller engineering effort required to manually annotate the training images. We manually made coarse annotations of the iris and periocular regions (122K images from the visible (VIS) spectrum and 38K images from the near-infrared (NIR) spectrum). The iris annotations in the NIR databases were generated semi-automatically by first applying an iris segmentation CNN and then performing a manual inspection. These annotations were made for 11 well-known public databases (3 NIR and 8 VIS) designed for the iris-based recognition problem and are publicly available to the research community. Experimenting our proposal on these databases, we highlight two results. First, the Faster R-CNN + Feature Pyramid Network (FPN) model reported an Intersection over Union (IoU) higher than YOLOv2 (91.86% vs 85.30%). Second, the detection of the iris and periocular regions being performed simultaneously is as accurate as performed separately, but with a lower computational cost, i.e., two tasks were carried out at the cost of one.

Sex-Prediction from Periocular Images across Multiple Sensors and Spectra

Abstract: In this paper, we provide a comprehensive analysis of periocular-based sex-prediction (commonly referred to as gender classification) using state-of-the-art machine learning techniques. In order to reflect a more challenging scenario where periocular images are likely to be obtained from an unknown source, i.e. sensor, convolutional neural networks are trained on fused sets composed of several near-infrared (NIR) and visible wavelength (VW) image databases. In a cross-sensor scenario within each spectrum an average classification accuracy of approximately 85\% is achieved. When sex-prediction is performed across spectra an average classification accuracy of about 82\% is obtained. Finally, a multi-spectral sex-prediction yields a classification accuracy of 83\% on average. Compared to proposed works, obtained results provide a more realistic estimation of the feasibility to predict a subject's sex from the periocular region.

Nighttime periocular recognition at long standoffs with deep learned features

Abstract: The periocular region is considered as a relatively new modality of biometrics and serves as a substitute solution for face recognition with occlusion. Moreover, many application scenarios occur at nighttime, such as nighttime surveillance. To address this problem, we study the topic of periocular recognition at nighttime using the infrared spectrum. Utilizing a simplified version of DeepFace, a convolutional neural networks designed for face recognition, we investigate nighttime periocular recognition at both short and long standoffs, namely 1.5 m, 50 m and 106 m. A subband of the active infrared spectrum { near-infrared (NIR) { is involved. During generation of the periocular dataset, preprocessing is conducted on the original face images, including alignment, cropping and intensity conversion. The verification results of the periocular region using DeepFace are compared with the results of two conventional methods { LBP and PCA. Experiments have shown that the DeepFace algorithm performs fairly well (with GAR over 90% at FAR=0.1%) using the periocular region as a modality even at nighttime. The framework also shows superiority to both LBP and PCA in all cases of different light wavelengths and standoffs.