Search

Automated facial recognition methods have become widely used as a way to ascertain the identity of individuals. Yet the methods by which facial recognition technologies (FRT) operate – the machinic performance of the perception of the human face – are often invisible to those under their gaze. This article investigates the machinic perception of the face through an FRT method known as eigenface, in order to both reveal and problematize the ways of seeing that underlie it. As part of its algorithmic processes, eigenface produces an image. This image can be understood as a portrait of machine recognition, making visible the processes through which the algorithm performs recognition and ‘sees’ the human face. The eigenface portrait reveals a way of seeing that is based on statistical processes of pattern recognition. An analogue antecedent of this application of statistics to the recognition of facial images can be found in the composite portrait. Through a dialectical discussion of composite portraiture in multiple disciplinary fields ranging from sociology to philosophy and the visual arts, this article experiments with providing a cultural and social translation of machine processes of visual perception. The discussion shifts the focus of enquiry towards the aesthetics of the algorithmic process in order to provide an entry point for critique and a possible reimagination on algorithmic knowledge production. (Less)

Technology has developed so well even to the point where machines can recognize our faces. There many APIs and algorithms out there that can recognize human faces. These APIs or algorithms is called Artificial Intelligence or AI. As modern civilian life in modern society, AI has been used so many times and developed into something more. This time, we proposed an idea to recognize motions and emotions on human faces to detect their stress levels. Companies nowadays need to have something that can know how their employees are feeling or think about their work. This program works by taking an image input from mounted cameras on the office’s monitor. However, before that, the program must have been trained first to recognize faces and emotions. For this, the eigenfaces algorithm is used. After the input is received, then the program will process the image received and extract the features from the images such as eyes, nose, mouth, etc. to detect if there are faces and also the emotions of the face. After that, the features extracted will be matched with the features in the database to give an output of the percentage of stress.

This work aims to present face recognition solution using eigenfaces which can be used for various applications like the online attendance system, access control and others. The patterns in human faces have been extracted using the Principal Component Analysis (PCA) and the extracted eigenfaces, which are the eigenvectors of the covariance matrix, represent the extracted features. Euclidean distance based classifier has been utilized which compares the Euclidean distance between the test image and the face images in the training set to ascertain the class to which the test image should belong. Two varying datasets are used in this work with first containing selected facial images from that AT&T or the Olivetti Research Laboratory facial database while second set has been generated locally at DHA Suffa University, Karachi. Seventy percent of the images have been used for training while the remaining thirty percent images have been used for evaluating the classifier. The implemented algorithm is developed in MATLAB and ensures an overall efficiency of around ninety percent for slight variations in facial expressions and postures.

Data security, complete system control, and missed storage and computing opportunities in personal portable devices are some of the major limitations of the centralized cloud environment. Among these limitations, security is a prime concern due to potential unauthorized access to private data. Biometrics, in particular, is considered sensitive data, and its usage is subject to the privacy protection law. To address this issue, a multimodal authentication system using encrypted biometrics for the edge-centric cloud environment is proposed in this study. Personal portable devices are utilized for encrypting biometrics in the proposed system, which optimizes the use of resources and tackles another limitation of the cloud environment. Biometrics is encrypted using a new method. In the proposed system, the edges transmit the encrypted speech and face for processing in the cloud. The cloud then decrypts the biometrics and performs authentication to confirm the identity of an individual. The model for speech authentication is based on two types of features, namely, Mel-frequency cepstral coefficients and perceptual linear prediction coefficients. The model for face authentication is implemented by determining the eigenfaces. The final decision about the identity of a user is based on majority voting. Experimental results show that the new encryption method can reliably hide the identity of an individual and accurately decrypt the biometrics, which is vital for errorless authentication.

Personal identification or verification is a very common requirement in modern society specially to access restricted area or resources. Biometric identification specially faces identification or recognition in a controlled or an uncontrolled scenario has become one of the most important and challenging area of research. Images often are represented as high-dimensional vectors or arrays. Operating directly on these vectors would lead to high computational costs and storage demands. Also working directly with raw data is difficult, challenging or even impossible sometimes. Dimensionality reduction has become a necessity for pre-processing data, representation and classification. It aims to represent data in a low-dimensional space that captures the intrinsic nature of the data. In this article we have applied a Supervised distance preserving projection (SDPP) technique, Semidefinite Least Square SDPP (SLS-SDPP), we have proposed recently to reduce the dimension of face image data. Numerical experiments conducted on very well-known face image data sets both on gallery images and blurred images of various level demonstrate that the performance of SLS-SDPP is promising in comparison to two leading approach Eigenface and Fisherface.

The fusion of multiple features is important for achieving state-of-the-art face recognition results. This has been proven in both traditional and deep learning approaches. Existing feature fusion methods either reduce the dimensionality of each feature first and then concatenate all low-dimensional feature vectors, named as DR-Cat, or the vice versa, named as Cat-DR. However, DR-Cat ignores the correlation information between different features which is useful for classification. In Cat-DR, on the other hand, the correlation information estimated from the training data may not be reliable especially when the number of training samples is limited. We propose a covariance matrix regularization (CMR) technique to solve problems of DR-Cat and Cat-DR. It works by assigning weights to cross-feature covariances in the covariance matrix of training data. Thus the feature correlation estimated from training data is regularized before being used to train the feature fusion model. The proposed CMR is applied to 4 feature fusion schemes: fusion of pixel values from 3 color channels, fusion of LBP features from 3 color channels, fusion of pixel values and LBP features from a single color channel, and fusion of CNN features extracted by 2 deep models. Extensive experiments of face recognition and verification are conducted on databases including MultiPIE, Georgia Tech, AR and LFW. Results show that the proposed CMR technique significantly and consistently outperforms the best single feature, DR-Cat and Cat-DR. NRF (Natl Research Foundation, S’pore) MOE (Min. of Education, S’pore) Accepted version

Face recognition is an algorithm that is capable of identifying or verifying a query face from multiple faces in the enrollment database. It poses a challenging problem in the field of image analysis and computer vision, especially for applications that deal with video sequences, face re-identification, or operate on intensity images and require fast processing. In this work, we introduce a high speed face recognition technique along with a high speed FPGA implementation. It uses a new similarity measure to estimate the distance between the query face and each of the database face images. The distance metric is the sum of the standard deviations between multiple histograms, which are calculated from each row of the query and database images. The lowest distance score refers to the database face that matches the query. The proposed technique is independent from the ambient illumination and outperforms the well-known face recognition algorithm "Eigenfaces" (it performs the face recognition 16 x faster when both algorithms run on the same platform). Furthermore, we exploit data parallelism in our proposed algorithm to design a hardware accelerator and to implement it on an FPGA prototyping board. The results show 10x execution time improvement in comparison to the software version.

Penelitian ini membahas tentang teknologi biometrik yang digunakan untuk mengenali pola fisik dari seseorang sehingga dapat teridentifikasi. Focus biometrik yang akan diteliti adalah deteksi wajah. Hal ini dilakukan karena melihat kurang optimalnya pengenalan biometrik pada sidik jari yang rentan tidak terdeteksi akibat luka atau perubahan struktur kulit seseorang. Teknologi biometrik wajah ini dapat digunakan untuk mengenali seseorang dalam melakukan perekaman data seperti pengelolaan biodata pegawai, masyarakat dan absensi. Untuk membangun teknologi biometrik tersebut maka dibutuhkan sebuah algoritma yang mampu melakukan pengenalan pola yaitu Eigenface. Algoritma tersebut telah teruji dan banyak digunakan pada penelitian sebelumnya dalam pengenalan pola, sehingga sangat tepat untuk digunakan dalam penelitian ini. Proses pengenalan wajah dimulai dengan mengambil citra wajah sehingga akan memperoleh nilai RGB, kemudian melakukan konversi dari RGB ke grayscale. Setelah proses konversi selesai maka selanjutnya mengubah citra tersebut ke dalam bentuk matriks dan menghitung nilai Flatvektor sehingga nantinya akan dicari nilai kedekatan dari data citra yang direkam. Dari hasil penelitian yang dilakukan maka diperoleh hasil bahwa algoritma Eigenface telah berhasil melakukan pengenalan wajah dengan akurasi 100% dari 10 data uji. Pengujian akurasi dilakukan dengan insitas cahaya 10 sampai dengan 1000 Lux, jarak antara 30 sampai dengan 200 cm dan gerak objek diatas 4 detik

Automatic facial recognition is fast becoming a reliable method for identifying individuals. Due to its reliability and unobtrusive nature facial recognition has been widely deployed in law enforcement and civilian application. Recent implementations of facial recognition systems on public cloud computing infrastructures have raised strong concerns regarding an individual's privacy. In this paper, we propose and implement a novel approach for facial recognition in the encrypted domain. This allows for facial recognition to be performed without revealing the actual image unnecessarily as the features stay encrypted at all times. Our proposed system exploits the homomorphic properties of the Paillier cryptosystem and performs Euclidean distance calculations using encrypted data. We propose to represent the images using a radial Local Ternary Pattern approach where a higher than proposed radius is used to extract the image features. Our proposed system has been evaluated using two publicly available datasets and has also been compared against the previously used eigenface approach in the encrypted domain and the obtained results justify the feasibility of the proposed system. © 2021 Elsevier Ltd

Face recognition is a technique that can be used to distinguish the characteristic facial patterns of a person. A very influencing factor in the facial recognition process is image quality, so it can affect the level of accuracy. Improving the accuracy of low-quality facial recognition can be done by processing the image for its feature extraction using Two Directional Matrix on 2D-PCA. From the extraction process, the Eigenfaces value is then generated to be classified. Image classification is done by using Euclidean Distance. Furthermore, the accuracy results between image recognition with or without Two Directional Matrix will be compared. The data used were the AT&T face of database and database of Essex. Of the 18 tests carried out on the Two Directional Matrix method, it was proven to improve facial recognition accuracy by as many as 12 trials. In 5 other experiments the accuracy decreased and 1 experiment was unable to increase or decrease facial recognition accuracy. The highest accuracy result in the experiment using AT&T was 98.25%, while in Essex it was 97.27%. Suggestions for further research by conducting experiments with more diverse dataset ratios in order to get better low-quality image accuracy results.

A framework to detect, identify and track individuals is proposed.Agents communication language complies with FIPA standards using cloud storage.Manager, face detector and face tracker agents interact by contract net protocol.Face detection uses Scale Invariant Feature Transform/Speeded Up Robust Features.The proposed framework tracks individuals in unconstrained environments. Agents with intelligent perceptual capabilities are considered state-of-the-art in advanced intelligent systems. In addition, a multiagent system is considered an enabling technology for applications that rely on distributed and parallel processing, including data, information and knowledge in complex computing environments. With the aim of creating advanced intelligent systems with visual perception, this paper presents an agent-based framework to individual tracking in unconstrained environments. The framework has three types of agents that interact using the Contract Net Protocol. The face detector and tracker agents perform fully automatic single-sample face recognition using the Viola-Jones and the Scale Invariant Feature Transform/Speeded Up Robust Features algorithms. The experimental results show that the framework adequately recognizes and tracks individuals in unconstrained environments, indicating the path the individuals have taken and the time they spent in the field of view of the surveillance agents. Some of the open source framework advantages are the distribution in heterogeneous infrastructures, the expansion with new agents using different face recognition algorithms (e.g., Eigenfaces), and the individual tracking logs that can be used in different ways, e.g., improve security in surveillance areas such as automated teller machines, self-paying kiosks, movie box offices, and malls.

This research is motivated by the condition of children with autism who require continuous monitoring without a parent accompany, and certain therapies who is capable to reduce repetitive behaviors and increase the concentration. Music therapy is one of the treatment which able to perform it. The implementation of this research is located in a special schools with autism, the perform is proper at certain times and requires tools for music therapy. Therefore, the integrated device was made to monitor the activities of children with autism, as well as providing music therapy automatic face recognition based on Eigenface method.This device perform when the children with autism under certain conditions (e.g. crying, the teacher would not work orders, etc.), then by capturing the the facial image, the system will process to compare the similarity with the existing database. Then, the shortest distance of euclidean is chosen. If the captured of facial image is similar to the one existing database, then the music is performed as music therapy for children with autism. The results of this system, indicates that the child's responses become more calm, easy concentration, and the repetitive attitude is reduced. While the accuracy of the system achieves by 80% (compare with the old and new database) and 20%. (without new database). ©2017 JNSMR UIN Walisongo. All rights reserved.

Sparse representation-based classification (SRC) method has gained great success in face recognition due to its encouraging and impressive performance. However, in SRC the data used to train or test are usually corrupted, and hence the performance is affected. This paper proposes a robust face recognition approach by means of learning a class-specific dictionary and a projection matrix. Firstly, the training data are decomposed into class-specific dictionary, non-class-specific dictionary, and sparse error matrix. Secondly, in order to correct the corrupted test data, the data are projected onto their corresponding underlying subspace, and a projection matrix between the original training data and the class-specific dictionary is learned. Then, the features of the class-specific dictionary and the corrected test data are extracted by using Eigenface method. Finally, the SRC is performed to classify. Extensive experiments conducted on publicly available data sets show that the proposed algorithm performs better than some state-of-the-art methods.

The assumption that the number of training samples is less than the number of pixels in a face image is essential for conventional eigenface-based face recognition. But recently, it has become impractical for massive face image collections. A parallel processing method using distributed eigenfaces is presented. A massive face image set was divided into a bunch of small subsets that satisfied the assumption of conventional approaches. Eigenfaces were extracted from the subsets and stored in a cloud system. Face recognition was performed by parallel processing using the distributed eigenfaces in the cloud system. A face recognition system was implemented in the Hadoop system. Various experiments were performed to test the validity of the distributed eigenface-based approach. The experimental results show that, compared to conventional methods, the implemented distributed face recognition system worked well for large datasets without significant performance degradation.

This paper addresses the applicability of multi-class Logical Analysis of Data (LAD) as a face recognition technique (FRT). This new classification technique has already been applied in the field of biomedical and mechanical engineering as a diagnostic technique; however, it has never been used in the face recognition literature. We explore how Eigenfaces and Fisherfaces merged to multi-class LAD can be leveraged as a proposed FRT, and how it might be useful compared to other approaches. The aim is to build a single multi-class LAD decision model that recognizes images of the face of different persons. We show that our proposed FRT can effectively deal with multiple changes in the pose and facial expression, which constitute critical challenges in the literature. Comparisons are made both from analytical and practical point of views. The proposed model improves the classification of Eigenfaces and Fisherfaces with minimum error rate.

Authentication is the process of verifying one’s identity, and one of its implementation is in taking attendances in university’s lectures. Attendance taking is a very important matter to every academic institution as a way to examine students’ performance. Signature based attendance taking can be manipulated. Therefore it has problems in verifying the attendance validity. In this final project, a real time eigenface based face recognition is implemented in an application to do attendance taking. The input face image is captured using a webcam. The application itself is built in C#, utilizing EmguCV library. The application is developed using Visual Studio 2015. Face detection is done with Viola-Jones algorithm. The eigenface method is used to do facial recognition on the detected face image. In this final project, a total of 8 testings are done in different conditions. From the testings, it is found that this application can recognize face images with accuracy as high as 90% and as low as 6.67%. This solution can be used as an alternative for real-time attendance taking in an environment with 170 lux light intensity, webcam resolution of 320 x 240 pixel, and the subject standing 1 meter away while not wearing spectacles. The average recognition time is 0.18125 ms.

This study develops an example-based face hallucination system based on a novel two-dimensional direct combined model (2DDCM) approach. The 2DDCM model combines the low-resolution and high-resolution pairwise images in the training set in a combined (or concatenated) matrix form in order to better preserve the correlation between the two images during the system learning process. Notably, the images processed by the 2DDCM model have the form of two-dimensional (2D) matrices rather than one-dimensional (1D) vectors, and hence the facial geometry features in the vertical and horizontal directions can be more reliably extracted. The proposed hallucination system comprises two 2DDCM-based modules, namely a global module for global facial structure reconstruction and a local module for facial texture detail compensation. In implementing the local module, a 2DDCM-based bi-directional transformation method is adopted to identify the detailed facial textures which are lost in the global synthesis process. The experimental results show that the synthesized results obtained using the proposed 2DDCM framework are in good quantitative agreement with the ground-truth images. Moreover, the proposed framework demonstrates the ability to synthesize high-resolution facial images given only a small number of training pairs, even when the facial features, alignment and appearance of the testing image differ from those of the original training set. Finally, the 2DDCM representation ensures that the synthesized results better preserve the subject-specific characteristics of the input facial image, and therefore improves the performance of downstream applications such as automatic face recognition.

In this modern era, social media facilitates us in communicating with the people across the world. Sharing of photos on social networks is due to their addictive interest in receiving the likes from other users in the network to gain popularity. Users on social network upload nearly 1.8 billion photos every day. Malicious users try to gather the publicly available photos on social networks and use it for creating bogus accounts. To identify those anomalous users, the photos shared are collected and processed by the social network manager to classify the original person from the fake one. As there are billions of users in each social network, there are an enormous amount of photo uploads which leads to the problem of scalability, slower processing performance and execution speed. The primary objective of this paper is to identify the similar image for a given a query image on a large set of image datasets crawled from online social networks through the Internet. For handling the scalability incurred from large image data sets, the image matching computation is implemented in distributed computing Map Reduce framework. The face recognition involves supervised machine learning approach employing Computer vision algorithms, namely Fisher face, Eigenface and Local Binary Pattern Histogram (LBPH). The similarity functions are used to calculate the distance between the query image and image sequence in the Hadoop file system. Experiments use the trained data sets to find the least similarity measure. The results obtained show that LBPH provides better and accurate matching results compared to other two face recognition approaches.

Identifying and identifying a person's face is one of the challenging issues in computer applications. The main part is the facial recognition of the whole image. Problems include sensitivity to the exposure conditions of the input image. Here is a way to detect human faces in color images based on the edge and color of the skin of the color image by setting the appropriate thresholds. First, image enhancement is performed, especially if the image is obtained from an infinite lighting condition. Then the skin was digested in the laboratory environment. The edges of the image are combined with the tone of the skin color to separate all areas other than the face. For this purpose, skin color detection methods have been used to detect faces. The Eigenface method is also known as the PCA method. This process has been tested in pictures and samples for various samples and has yielded good results. Due to the increasing instances of identity theft and the occurrence of terrorism in the past years, recognizing and identifying a person can determine the unity of a human being with another person's face. For facial recognition of an online monitoring system or an offline image, the main component to be identified is the skin areas. The skin color has been proven to be a powerful and useful indication for face detection, placement, and follow-up. Face detection is simply a facial image for determining the input given, regardless of its size, position, and background. The current evolution of computer technology has increased in this era.

As cost-effective and relatively accurate models for behavior classification, automatic facial expression analysis has the potential to be applied to multiple disciplines. The current research deals with real-time classification of evoked emotions in children. Reason behind analyzing children behavior is that they possess immature cognitive abilities compared to adults, they are more likely to be influenced by external stimuli and they are less likely to pose and hide expression. This research work uses a three-phase classification model to analyze real time captured emotion varying children facial expressions. Two classifiers K-Nearest Neighbor and SVM were used for classification. The classification model was tested using our own real time recorded children Facial Expression (CFE).

Based on previous research that has been conducted by other researchers about the Eigenface and Fisherface algorithms, where face image recognition uses eigenfaces with various conditions of variation from expression to face assessors has success rate until 75%. Adaboost face recognition, Eigenface PCA and MySQL produce 80% of various different conditions. Face recognition with fisherfaces by 73 face trials of different expressions and different positions, 93% success. From the previous research, the author wanted to know which algorithm has a level of speed in terms of the accuracy of time, distance and light sensitivity in the face recognition process. From the trials of the two algorithms, it produces success percentages and accuracy charts. However, this research shows us how to implement of Ficherface and Eigenface Algorithm by experimental, then show us the time process of Fisherface Algorithm and time process of Eigenface Algorithm for face recognition. The results of this study are Time Performance The Fisherface algorithm is faster than the eigenface algorithm and the accuracy of the Eigenface Algorithm is 96% and the accuracy of the fisherface algorithm is 97%.

Face detection and picture or video recognition is a popular subject of research on biometrics. Face recognition in a real-time setting has an exciting area and a rapidly growing challenge. Framework for the use of face recognition application authentication. This proposes the PCA (Principal Component Analysis) facial recognition system. The key component analysis (PCA) is a statistical method under the broad heading of factor analysis. The aim of the PCA is to reduce the large amount of data storage to the size of the feature space that is required to represent the data economically. The wide 1-D pixel vector made of the 2-D face picture in compact main elements of the space function is designed for facial recognition by the PCA. This is called a projection of self-space. The proper space is determined with the identification of the covariance matrix’s own vectors, which are centered on a collection of fingerprint images. I build a camera-based real-time face recognition system and set an algorithm by developing programming on OpenCV, Haar Cascade, Eigenface, Fisher Face, LBPH, and Python.

Principal Component Analysis (PCA) has been successfully used for many applications, including ear recognition. This paper presents a 2D Wavelet based Multi-Band PCA (2DWMBPCA) method, inspired by PCA based techniques for multispectral and hyperspectral images, which have shown a significantly higher performance to that of standard PCA. The proposed method performs 2D non-decimated wavelet transform on the input image dividing the image into its subbands. It then splits each resulting subband into a number of bands evenly based on the coefficient values. Standard PCA is then applied on each resulting set of bands to extract the subbands eigenvectors, which are used as features for matching. Experimental results on images of two benchmark ear image datasets show that the proposed 2D-WMBPCA significantly outperforms both the standard PCA method and the eigenfaces method.