Automatic Face Recognition System Research Articles

Cross-resolution face identification using deep-convolutional neural network

Low resolution (LR) and very low resolution (VLR) face images captured by surveillance cameras make automatic face recognition (AFR) a challenging task. The performance of an automatic face recognition system (AFRS) degrades when these types of face images are compared with high resolution (HR) gallery images. This paper has presented a face identification system called Cross-Resolution Face Identification System, to address this issue. It makes use of the Deep Convolutional Neural Network (DCNN) having different pooling operations to extract resolution robust features from high resolution, low resolution and very low resolution face images. The proposed system is evaluated on four face databases, namely, the ORL, the extended Yale face B, the LFW, and the Georgia Tech under three cross resolution environmental conditions based on resolution of probe images (i.e., high resolution to high resolution, low resolution to high resolution, and very low resolution to high resolution).The experimental outcomes exhibit the effectiveness of the proposed face identification system.

Iris pattern recognition as an infallible mean of identity establishment in forensic casework

Biometrics is a mean of forensic recognition based on me as ureme nts an d calculat ions o f human bo dy or biological features. Human body features. Biometric identification systems may make use of iris, retina, fingerprints, face, ear, gait parameters, facial physiognomy, photogrammetric or geometry features, etc. for recognition and hence identification pursuits. The high security areas like airports, nuclear reactors, thermal or hydro-electric power plants, legislative buildings, defense establishments etc., require secured entry of authorized persons. Biometric tools of recognition like automatic face recognition system can play a crucial role in recognizing illegal intruders to such areas. Iris is an excellent trait to be used in biometric system for recognition of an individual due to its individualistic, unique, universal and inimitable features. In this article, the current national and international status of use of iris patterns forforensic recognition, its advantages and limitations, and future prospects have been presented.

Forensic face recognition based on KDE and evidence theory

Forensic face recognition (FFR) has been studied in recent years in forensic science. Given an automatic face recognition system, output scores of the system are used to describe the similarity of face image pairs, but not suitable for forensics. In this study, a score-mapping model based on kernel density estimation (KDE) and evidence theory is proposed. First, KDE was used to generate probability density function (PDF) for each dimensional feature vector of face image pairs. Then, the PDFs could be utilized to determine separately the basic probability assignment (BPA) of supporting the prosecution hypothesis and the defence hypothesis. Finally, the BPAs of each feature were combined by Dempster’s rule to get the final BPA, which reflects the strength of evidence support. The experimental results demonstrate that compared with the classic KDE-based likelihood ratio method, the proposed method has a better performance in terms of accuracy, sensitivity and specificity.

Automated Artifact Retouching in Morphed Images With Attention Maps

Morphing attack is an important security threat for automatic face recognition systems. High-quality morphed images, i.e. images without significant visual artifacts such as ghosts, noise, and blurring, exhibit higher chances of success, being able to fool both human examiners and commercial face verification algorithms. Therefore, the availability of large sets of high-quality morphs is fundamental for training and testing robust morphing attack detection algorithms. However, producing a high-quality morphed image is an expensive and time-consuming task since manual post-processing is generally required to remove the typical artifacts generated by landmark-based morphing techniques. This work describes an approach based on the Conditional Generative Adversarial Network paradigm for automated morphing artifact retouching and the use of Attention Maps to guide the generation process and limit the retouch to specific areas. In order to work with high-resolution images, the framework is applied on different facial crops, which, once processed and retouched, are accurately blended to reconstruct the whole morphed face. Specifically, we focus on four different squared face regions, i.e. the right and left eyes, the nose, and the mouth, that are frequently affected by artifacts. Several qualitative and quantitative experimental evaluations have been conducted to confirm the effectiveness of the proposal in terms of, among the others, pixel-wise metrics, identity preservation, and human observer analysis. Results confirm the feasibility and the accuracy of the proposed framework.

Clinical application of an automatic facial recognition system based on deep learning for diagnosis of Turner syndrome.

Automated facial recognition technology based on deep learning has achieved high accuracy in diagnosing various endocrine diseases and genetic syndromes. This study attempts to establish a facial diagnostic system for Turner syndrome (TS) based on deep convolutional neural networks. Photographs of 207 TS patients and 1074 female controls were collected from July 2016 to April 2019. Finally, 170 patients diagnosed with TS and 1053 female controls were included. Deep convolutional neural networks were used to develop the facial diagnostic system. A prospective study, which included two TS patients and 35 controls, was conducted to test the efficacy in the real clinical setting. The average areas under the curve (AUCs) in three different scenarios were 0.9540 ± 0.0223, 0.9662 ± 0.0108 and 0.9557 ± 0.0119, separately. The average sensitivity and specificity of the prospective study were 96.7% and 97.0%, respectively. The facial diagnostic system achieved high accuracy. Prospective study results demonstrated the application value of this system, which is promising in the screening of Turner syndrome.

Human face recognition methods based on principle component analysis (PCA), wavelet and support vector machine (SVM) : a comparative study

<p class="JESTECAbstract">Human face Recognition systems are increasingly gaining more importance and can be utilized throughout many applications like video surveillance, Security, human-computer intelligent interaction, etc. this paper presents performance comparison between three feature extraction techniques for an automatic face recognition system. In the first step, we benefit from wavelet Transforms, Principal Component Analysis (PCA) and combining Wavelet with PCA as feature extracting methods. After feature vectors generation, linear and nonlinear Support Vector Machines (SVM) are usually used for implementing the classification or recognition step. These methods are compared on accuracy in an ORL database for face recognition applications including 400 images of 40 people.</p>

3D face reconstruction from mugshots: Application to arbitrary view face recognition

Mugshots while routinely acquired by law enforcement agencies are under utilized by automated face recognition systems. In this paper, we propose a regression based approach to reconstruct textured full 3D face models from multi-view mugshot images. Using landmarks from the input frontal and profile mugshots of a subject, our method reconstructs his⧹her 3D face shape via either linear or nonlinear regressors. The texture of the mugshot images is mapped to the reconstructed 3D face shape via an efficient seamless texture recovery scheme. Compared with existing 3D face reconstruction methods, the proposed method more effectively utilizes the three-view mugshot face images collected during booking. The reconstructed 3D faces are used to generate realistic multi-view face images to enlarge the gallery and facilitate arbitrary view face recognition. Evaluation experiments have been done on BFM and Bosphorus databases in terms of reconstruction accuracy, and on Multi-PIE and Color FERET databases in terms of recognition accuracy. The results show that the proposed method can reduce the 3D face reconstruction error of the best competitive method from 2.31 mm to 1.88 mm, and improve the recognition accuracy of state-of-the-art deep learning based face matchers by as much as ~4% on Multi-PIE and ~2% on Color FERET despite the high baseline set by them.

A Novel Image Description With the Stretched Natural Vector Method: Application to Face Recognition

Nowadays there is growing research interest in designing high performance algorithms for automatic facial recognition systems, and an efficient computational approach is required. Accurate face recognition, however, is difficult due to facial complexity. In this paper, we propose a novel and efficient facial image representation named the Stretched Natural Vector (SNV) method which is defined on the intensity values in a grayscale image matrix, where each entry in an intensity matrix records the level of gray at a single pixel in a m×n array. We prove that the SNV defined in this context can distinguish photo matrices in strict one-to-one fashion. This is to say it is theoretically possible to fully recover a grayscale image matrix from the corresponding complete SNV. Experiments on a number of datasets demonstrate that our truncated SNV method compares favorably both in recognition accuracy and efficiency (measured in wall-clock time) against “Full-Pixel” algorithm, Principal Component Analysis (PCA) method, and even its widely used variants - two dimensional PCA (2DPCA) method and two dimensional Euler PCA (2D-EPCA) method.

Face Identification Based on Discrete Wavelet Transform and Neural Networks

The subject paper presents implementation of a new automatic face recognition system. To formulate an automated framework for the recognition of human faces is a highly challenging endeavor. The face identification problem is particularly very crucial in the context of today’s rapid emergence of technological advancements with ever expansive requirements. It has also significant relevance in the related engineering disciplines of computer graphics, pattern recognition, psychology, image processing and artificial neural networks. This paper proposes a side-view face authentication approach based on discrete wavelet transform and artificial neural networks for the solution of the problem. A subset determination strategy that expands on the number of training samples and permits protection of the global information is discussed. The authentication technique involves image profile extraction, decomposition of the wavelets, splitting of the subsets and finally neural network verification. The procedure exploits the localization property of the wavelets in both the frequency and spatial domains, while maintaining the generalized properties of the neural networks. The realization strategy of the methodology was executed using MATLAB, demonstrating that the performance of the technique is quite satisfactory.

Digital Attendance Register using Raspberry-Pi

Nowadays marking the attendance of the students is a very important work for every staff in every class. The staff has to mark the attendance by manually by seeing the student faces. After verifying the student’s, they have to cross check and the register will be submitted to the higher officials by manually. It consumes a lot of time for the processing. Currently, the proposed system is useful for colleges to mark the attendance automatically without any manual dependent. Generally, face is the most identified part for every human. Hence, the proposed work is automatic face recognition system by using raspberry pi and Open CV software. This work involves five methods namely, face detection, face preprocessing, face training, face recognition and attendance marking. The attendance will be marked in the IOT page and used an LCD display for the details of that student.

Face Authenticity: An Overview of Face Manipulation Generation, Detection and Recognition

In recent years, there has been an exponential increase in photo and video manipulation by easy-to-use editing tools (e.g., Photoshop). Especially, ‘face digital manipulations’ (e.g., face swapping) is a critical issue for automated face recognition systems (AFRSs) as it detrimentally effects the AFRS’ performance. Also, the advent of powerful deep learning methods has led to realistic face sample generation and manipulation. Despite recent advances in face manipulation detection techniques, manipulations-aware AFRSs and face synthetic sample/manipulation generation, detecting sophisticated face manipulations is still a challenge to human examiners and existing technologies. Devising more effective universal face manipulation detectors and manipulations-aware AFRSs will immensely improve the trust in biometric applications and digital communications. This paper presents an overview of the recent technologies on face manipulation generation, detection, recognition, and databases. Also, potential future research directions and challenges are discussed.

Analysis of Different M-Band Wavelet Filters for Face Recognition using Nearest Neighbor Classifier

<span>Face recognition system is one of the most interesting studied topics in computer vision for past two decades. Among the other popular biometrics such as the retina, fingerprint, and iris recognition systems, the face recognition is capable of recognizing the uncooperative samples in a non-intrusive manner. Also, it can be applied to many applications of surveillance security, forensics, border control, digital entertainment where face recognition is used in most. In the proposed system an automatic face recognition system is discussed. The proposed recognition system is based on the Dual-Tree M-Band Wavelet Transform (DTMBWT) transform algorithm and features obtained by varying the different filter in the DTMBWT transform. Then the different filter features are classified by means of the K-Nearest Neighbor (KNN) classifier for recognizing the face correctly. The implementation of the system is done by using the ORL face image database, and the performance metrics are calculated.</span>

Fusing magnitude and phase features with multiple face models for robust face recognition

High accuracy face recognition is of great importance for a wide variety of real-world applications. Although significant progress has been made in the last decades, fully automatic face recognition systems have not yet approached the goal of surpassing the human vision system, even in controlled conditions. In this paper, we propose an approach for robust face recognition by fusing two complementary features: one is Gabor magnitude of multiple scales and orientations and the other is Fourier phase encoded by spatial pyramid based local phase quantization (SPLPQ). To reduce the high dimensionality of both features, block-wise fisher discriminant analysis (BFDA) is applied and further combined by score-level fusion. Moreover, inspired by the biological cognitive mechanism, multiple face models are exploited to further boost the robustness of the proposed approach. We evaluate the proposed approach on three challenging databases, i.e., FRGC ver2.0, LFW, and CFW-p, that address two face classification scenarios, i.e., verification and identification. Experimental results consistently exhibit the complementarity of the two features and the performance boost gained by the multiple face models. The proposed approach achieved approximately 96% verification rate when FAR was 0.1% on FRGC ver2.0 Exp.4, impressively surpassing all the best known results.

Robust Singular Value Decomposition Algorithm for Unique Faces

It has been read and also seen by physical encounters that there found to be seven near resembling humans by appearance .Many a times one becomes confused with respect to identification of such near resembling faces when one encounters them. The recognition of familiar faces plays a fundamental role in our social interactions. Humans are able to identify reliably a large number of faces and psychologists are interested in understanding the perceptual and cognitive mechanisms at the base of the face recognition process. As it is needed that an automated face recognition system should be faces specific, it should effectively use features that discriminate a face from others by preferably amplifying distinctive characteristics of face. Face recognition has drawn wide attention from researchers in areas of machine learning, computer vision, pattern recognition, neural networks, access control, information security, law enforcement and surveillance, smart cards etc. The paper shows that the most resembling faces can be recognized by having a unique value per face under different variations. Certain image transformations, such as intensity negation, strange viewpoint changes, and changes in lighting direction can severely disrupt human face recognition. It has been said again and again by research scholars that SVD algorithm is not good enough to classify faces under large variations but this paper proves that the SVD algorithm is most robust algorithm and can be proved effective in identifying faces under large variations as applicable to unique faces. This paper works on these aspects and tries to recognize the unique faces by applying optimized SVD algorithm.

Extended StirTrace benchmarking of biometric and forensic qualities of morphed face images

Since its introduction in 2014, the face morphing forgery (FMF) attack has received significant attention from the biometric and media forensic research communities. The attack aims at creating artificially weakened templates which can be successfully matched against multiple persons. If successful, the attack has an immense impact on many biometric authentication scenarios including the application of electronic machine-readable travel document (eMRTD) at automated border control gates. We extend the StirTrace framework for benchmarking FMF attacks by adding five issues: a novel three-fold definition for the quality of morphed images, a novel FMF realisation (combined morphing), a post-processing operation to simulate the digital image format used in eMRTD (passport scaling 15 kB), an automated face recognition system (VGG face descriptor) as additional means for biometric quality assessment and two feature spaces for FMF detection (keypoint features and fusion of keypoint and Benford features) as additional means for forensic quality assessment. We show that the impact of StirTrace post-processing operations on the biometric quality of morphed face images is negligible except for two noise operators and passport scaling 15 kB, the impact on the forensic quality depends on the type of post-processing, and the new FMF realisation outperforms the previously considered ones.

A Comparative Evaluation of Dotted Raster-Stereography and Feature-Based Techniques for Automated Face Recognition

Automated face recognition systems are fast becoming a need for security-related applications. Development of a fool-proof and efficient face recognition system is a challenging domain for researchers. This paper presents comparative evaluation of two candidate techniques for automated face recognition application, viz. dotted Raster-stereography and feature-based system. The relevant performance parameters — accuracy, precision, sensitivity and specificity – measured for the two techniques using IPRL Database of images are reported. The results suggest that dotted Raster-stereography based face recognition system has better accuracy, precision, sensitivity and specificity, and hence is a preferred choice as compared with feature-based system for such sensitive applications where high face recognition accuracy is required. On the other hand, feature-based technique is faster in terms of the training and testing times required. Hence such applications where volume of face recognition work is large and high speed is required with some compromise in accuracy being acceptable then feature-based technique may also be the technique of choice.

Deep Learning for Understanding Faces: Machines May Be Just as Good, or Better, than Humans

Recent developments in deep convolutional neural networks (DCNNs) have shown impressive performance improvements on various object detection/recognition problems. This has been made possible due to the availability of large annotated data and a better understanding of the nonlinear mapping between images and class labels, as well as the affordability of powerful graphics processing units (GPUs). These developments in deep learning have also improved the capabilities of machines in understanding faces and automatically executing the tasks of face detection, pose estimation, landmark localization, and face recognition from unconstrained images and videos. In this article, we provide an overview of deep-learning methods used for face recognition. We discuss different modules involved in designing an automatic face recognition system and the role of deep learning for each of them. Some open issues regarding DCNNs for face recognition problems are then discussed. This article should prove valuable to scientists, engineers, and end users working in the fields of face recognition, security, visual surveillance, and biometrics.

Automated Skin Detection using Fuzzy Logic based Pixel Matching Algorithm

Objectives: Skin detection is the initial process of any automatic face recognition system. One of the fore most common approaches to detect the faces in facial pictures can be done by an automatic face detection method. In this paper, a new type of facial skin detection using the combination of Fuzzy and Pixel matching algorithms is proposed. Method/Statistical Analysis: The proposed system has three steps. Initially, the input facial color image is transformed in to a grey scale image and then, it is sharpened using a filter. Secondly, the pixels of the different types of facial grey scale images are computed. Finally, the computed pixels are compared with the original grey scale image based on the fuzzy rules. This process is done for the first pixel to the last pixel so that all the pixels which are present in the entire image can be included in the overall process. Findings: The algorithm is tested for different types of facial images for both the accuracy and the time taken to complete the skin detection process. The results of the experiments reveal that an accurate skin detection rate of 94.42% for the combination of fuzzy based pixel matching along with the dimension reduction methodology. Application/ Improvements: Future enhancements in this work are to propose an efficient algorithm to detect the facial images of group of faces present in a single image. Keywords: Dimension Reduction, Facial Image, Fuzzy Logic, Pixel Matching Algorithm, Skin Texture Analysis, Skin Detection

Regression-Based Automated Facial Image Quality Model

Nowadays, biometric technologies became reliable and widespread means of unobtrusive user authentication in a variety of real-world applications. The performance of an automated face recognition system has a strong relationship with the quality of the biometric samples. The facial samples can be affected by various quality factors, such as uneven illumination, low or high contrast, excessive brightness, blurriness, etc. In this article, the authors propose a quality estimation method based on linear regression analysis to characterize the relationship between different quality factors and the performance of a face recognition system. The regression model can predict the overall quality of a facial sample which reflects the effects of various quality factors on that sample. The weights assigned to the different quality factors by the linear regression model reflect the impact of those quality factors on the performance of the recognition system. Therefore, the prediction scores generated from the model is a strong indicator of the overall quality of the facial images. The authors evaluated the quality estimation model on the Extended Yale Database B. They also performed a study to understand which quality factors affect the face recognition the most.

Fuzzy reasoning model to improve face illumination invariance

Enhancing facial images captured under different lighting conditions is an important challenge and a crucial component in the automatic face recognition systems. This work tackles illumination variation challenge by proposing a new face image enhancement approach based on Fuzzy theory. The proposed Fuzzy reasoning model generates an adaptive enhancement which corrects and improves non-uniform illumination and low contrasts. The FRM approach has been assessed using four blind-reference image quality metrics supported by visual assessment. A comparison to six state-of-the-art methods has also been provided. Experiments are performed on four public data sets, namely Extended Yale-B, Mobio, FERET and Carnegie Mellon University Pose, Illumination, and Expression, showing very interesting results achieved by our approach.