International Journal of Biometrics (10 papers in press)
Fingerprint Indexing via BRIEF Minutia Descriptors
by Robert Pollak, Roland Richter
Abstract: We use BRIEF binary local image descriptors as minutia descriptors for indexing of biometric fingerprint databases. Tests with varying descriptor size and parametrization are performed on a proprietary database. Compared with the speed of an implementation of conventional minutiae matching, we find that BRIEF descriptors are fast enough for database indexing. The tested descriptors outperform two other image descriptors (LBP, HoG) from recent literature with respect to matching rates and average penetration rates.
Keywords: biometrics; fingerprints; indexing; BRIEF; image descriptors.
Mutually Reinforcing Motion-Pose Framework for Pose Invariant Action Recognition
by Manoj Ramanathan, Wei-Yun Yau, Nadia Magnenat Thalmann, Eam Khwang Teoh
Abstract: Action recognition from videos has many potential applications. However, there are many unresolved challenges, such as pose-invariant recognition, robustness to occlusion and others. In this paper, we propose to combine two most important characteristics of an action, motion of body parts and specific canonical poses observed in a novel mutually reinforcing framework to achieve pose-invariant action recognition. The proposed framework consists of two components. The first component is the pose-invariant feature extraction that captures the motion of body parts needed for action recognition. This forms the forward propagation motion path of our framework that recognizes an initial action using the extracted features. Each action is characterized by specific canonical stick poses. Given the training videos of an action, we propose an algorithm to extract a dictionary of normalized canonical stick poses. The second component of the framework is the pose hypothesis generation scheme that compares each of the extracted canonical sticks of the initial action recognized with the video frame to identify the most likely canonical stick pose. We use the identified canonical stick pose in the frame to improve the pose-invariant motion feature extraction. To capture the temporal dynamics of an action, we introduce temporal stick features computed using the stick poses obtained. This pose-based component acts as the feedback path in our framework. The combination of pose-invariant kinematic features from the forward and feedback paths together with the temporal stick features are used to recognize the action, thus forming a mutually reinforcing framework that repeats until the action recognition result converges. The proposed mutual reinforcement framework is capable of handling changes in posture of the person, occlusion and partial view-invariance. We perform experiments on several benchmark datasets which showed the performance of the proposed algorithm and its ability to handle pose variation and occlusion.
Keywords: Action recognition; pose-invariant motion feature; canonical stick poses; mutual reinforcement framework.
An Overlap based Human Gait Cycle Detection
by Sugandhi K., FARHA FATINA WAHID, Nikesh P, Raju G
Abstract: Identification of a person by his/her style of walking is referred as gait recognition. Gait is one among the biometric used for Human identification. In gait recognition, an inevitable step for accurate feature extraction is gait cycle detection. In this paper, a novel gait cycle detection algorithm based on the concept of overlap between legs during locomotion is proposed. To identify overlap, zero-crossing counts of silhouette frames as well as bottom halves of silhouette frames are considered. The efficiency of this algorithm is tested using normal walking sequence of subjects with 900 viewing angle from CASIA B as well as TUM-IITKGP human gait databases. The results obtained shows that gait cycle can be easily and efficiently detected with zero-crossing count of silhouette frames. Further zero-crossing counts taken from bottom halves of silhouette frames gives better performance.
Keywords: Gait; Gait cycle; Overlap; silhouette; zero-crossing.
Orthogonal rotation invariant features for iris and periocular recognition
by Bineet Kaur, Sukhwinder Singh, Jagdish Kumar
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 neighborhood 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.
Keywords: Biometrics; Iris Recognition; Orthogonal Moments; Periocular Recognition; Polar Complex Exponential Transform; Polar Cosine Transform; Polar Harmonic Transform; Polar Sine Transform; Zernike Moments.
Fake Fingerprint Liveness Detection Based on Micro and Macro Features
by Rohit Agarwal, Anand Singh Singh
Abstract: Finger print is the most hopeful biometric authentication that can specifically identify a person from their exclusive features. To make sure the difference between a real trait with a fake one constructed from dissimilar fabrication or reconstructed sample is a noteworthy hitch in fingerprint authentication. To keep an excellent point of protection, consistent spoofing detection tools are essential, preferably implemented as software modules. In the proposed approach, a novel software-based classification method is presented that can be used in fake fingerprint detection system to classify between fake and real one. The intention of the proposed system is to improve the security of biometric identification system, by accumulation liveness measurement in a rapid, comprehensible manner. For illustration while the statistical techniques are good for micro features but not well for macro. In this paper, we present a novel combination of local Haralick micro texture features with macro features derived from Neighborhood Gray-Tone Difference Matrix (NGTDM) to generate feature vector for training and testing fingerprint images. Combined extracted features of training and testing images are passed to Support Vector Machine for discriminating live and fake fingerprints. The proposed approach is experimented and validated on ATVS dataset and LivDet2011 dataset. The proposed approach has achieved good accuracy and very less error rate in comparison to the different previously studied techniques.
Keywords: Biometrics; Fingerprints; Liveness; Spoof; micro features; macro features;.
Real-Time Single-View Face Detection and Face Recognition based on Aggregate Channel Feature
by Michael George, Aswathy Sivan, Babita Roslind Jose, Jimson Mathew
Abstract: A single-view face detector and a novel face recognition method based on the Aggregate Channel Feature (ACF) that work at real-time speeds, suitable in a computing resource-constrained setting are presented in this work. The four stage tree based face detector is trained on a subset of the AFLW dataset. The face detection performance is analysed using the AFW dataset. The face recogniser uses ACF features along with classification algorithms, either SVM or MLP. The face recogniser is trained and tested on the GATech Face dataset. Our face detector displays comparable performance against the state of the art while working at 29.8 fps. The face recogniser achieves a level of performance that is competitive with other state of the art works. The effect of PCA based dimension reduction of ACF features on face recognition performance is also studied in this work.
Keywords: ACF; SVM; MLP; Face Recognition; Face Detection.
An Accurate Hand Based Multimodal Biometric Recognition System with Optimized Sum Rule for Higher Security Applications.
by Pallavi Deshpande, Prachi Mukherji, Anil Tavildar
Abstract: This paper presents a multimodal biometric recognition system using Palm print, Finger geometry and Dorsal Palm vein modalities. A specific image acquisition system is designed, fabricated and database of 150 users is created. DWT technique for features extraction is used for palm print and dorsal palm vein modalities. Performance analysis for individual modality is done using Receiver Operating Characteristics and accuracies of 98.775%, 98.45% and 97.60% are obtained respectively for PP, FG and DPV modalities. Further the multimodal system is proposed along with a novel basis for optimally choosing the weights. The score level fusion is done using these optimized weights. Testing, validation and bench marking of the algorithms are done using our own database, as well as the standard database available on the net. The proposed multimodal system gives enhanced accuracy of 99.80% with very low FAR level of 0.0001.
Keywords: Multimodal Biometric (MMB); Palm Print (PP); Dorsal Palm Vein (DPV); Finger Geometry (FG); False Acceptance Rate (FAR); Genuine Acceptance Rate (GAR); Receiver Operating Characteristics (ROC); Weights Optimization.
Study on Soft Behavioral Biometrics to Predict Consumers Interest Level Using Web Access Log
by Nobuyuki Nishiuchi, Seima Aoki
Abstract: Electric commerce (EC) market has grown fast and is also estimated to expand further in the future. Therefore, the value of the consumers access log on websites has been increased tremendously. Web analytics service also has been widely used, and the web analytics has become extremely effective. This paper presents a soft behavioral biometrics to predict the consumers interest level in a specific product using access log on websites. The experiments are conducted in a way where the subjects are asked to perform a shopping task on some websites. The comparative analysis is carried out between the interest level of one category product taken from the inquiry, and the access log during the purchasing process on websites. The results show that the behavioral patterns of the web searching and some parameters based on the access log are clearly different depending on the interest level. Moreover, based on the experiments data, an automatic classification of the interest level is tested using Support vector machine (SVM).
Keywords: soft biometrics; behavioral biometrics; consumer’s interest level; web access log; web analytics; purchasing process; electric commerce site; Support vector machine.
Recent trends of ROI segmentation in iris biometrics: a survey
by Ritesh Vyas, Tirupathiraju Kanumuri, Gyanendra Sheoran, Pawan Dubey
Abstract: Segmentation in iris biometrics deals with the localization of inner and outer boundaries of the iris and isolation of the region of interest (ROI) from the input eye image. The isolated ROI is further used to extract the meaningful features of iris for its effective representation. That is why accuracy of the segmentation module directly affects the overall accuracy in an iris recognition system. In view of this, the present study provides a comprehensive review of state-of-the-art methods on iris segmentation that were reported after 2011. Iris segmentation approaches based on eye images captured in both visible and near infrared illumination have been reviewed in this paper. The state-of-the-art iris segmentation approaches have been categorized into four broad classes, namely: Integro-Differential Operator (IDO) based approaches, Circular Hough Transform (CHT) based approaches, Deep Learning based approaches, and Miscellaneous approaches. The sole purpose of this survey is to deliver insights on ROI segmentation, which is a prominent step of iris recognition process, and to suggest prospective research directions to the readers.
Keywords: Iris biometrics; Region of Interest (ROI); iris segmentation; accuracy; near infrared (NIR); visible wavelength (VW).
A Common Convolutional Neural Network Model to Classify Plain, Rolled and Latent Fingerprints
by Asif Iqbal Khan
Abstract: Fingerprint classification helps in reducing the number of comparisons during the matching stage in automatic fingerprint identification system. In this study, a Convolutional Neural Network model is proposed for classification of plain, rolled and latent fingerprints. We first propose a new convolutional neural network model initialized with random weights and train the model on fingerprint images. Then we fine-tune two pre-trained convolutional neural network models on fingerprint images. Finally, we compare these three models: two pre-trained models and a custom convolutional neural network model initialized with random weights. We show that pre-trained models can achieve the state-of-the-art results on other similar tasks with no or little fine-tuning. We also show that training data size and depth of the network have a serious impact on the overall performance of deep networks. Dropout is used to enhance the performance of deep networks where the labeled training data is not of sufficient size. All the three models trained on NIST DB4 fingerprint and IIIT-D latent fingerprint databases report good accuracy. By only fine-tuning the pre-trained convolutional neural network model, we get the accuracy of 99%, easily out-performing the state-of-the-art
Keywords: Convolutional Neural Network (CNN/ConvNet); Deep Learning; Fingerprint classification; Latent Fingerprints.