Face Candidates Research Articles

Recognizability assessment of facial images for automated teller machine applications

Crimes related to automated teller machines (ATMs) have increased as a result of the recent popularity in the devices. One of the most practical approaches for preventing such crimes is the installation of cameras in ATMs to capture the facial images of users for follow-up criminal investigations. However, this approach is vulnerable in cases where a criminal's face is occluded. Therefore, this paper proposes a system which assesses the recognizability of facial images of ATM users to determine whether their faces are severely occluded. The proposed system uses a component-based face candidate generation and verification approach to handle various facial postures and acceptable partial occlusions. Element techniques are implemented via grayscale image-based methods which are robust against illumination conditions compared to skin color detection approach. The system architecture for achieving both high performance and cost-efficiency is proposed to make the system applicable to practical ATM environments. In the experiment, the feasibility of the proposed system was evaluated using a large-scale facial occlusion database consisting of 3168 image sequences including 21 facial occlusions, 8 illumination conditions, and 2 acquisition scenarios. Based on the results, we drew up the guidelines of recognizability assessment systems for ATM applications.

스마트 폰에서 취득한 얼굴영상에서 아이라인 검출에 관한 연구

In this paper, the extract method of eye and eye-line from picture of a person is proposed. Most of existing papers are to extract the position of eyeball but in this paper, by extracting not only the position of eyeball but also eye-line, it can be applied to the face application program variously. The experimental data of the input picture is a full face photograph taken by smart phone, basically the picture is limited to the face of one person and back ground can be taken from every where and no restriction of race. The proposed method is to extract face candidated area by using Harr Classifier and set up the candidate area of eye position from face candidate area. To extract high value from eye candidate area using dilate operation, and proposed the method to classify eye and eyelash by local thresholding of the picture. After that, using thresholding image from eyemapC that Hsu`s suggested, and separated the area with eye and without eye. Finally extract the contour of eye and detect eye-line using optimum ellipse estimation.

Complexity Reduced Face Detection Using Probability-Based Face Mask Prefiltering and Pixel-Based Hierarchical-Feature Adaboosting

The Adaboosting has attracted attention for its efficient face-detection performance. However, in the training process, the large number of possible Haar-like features in a standard sub-window becomes time consuming, which makes specific environment feature adaptation extremely difficult. This letter presents a two-stage hybrid face detection scheme using Probability-based Face Mask Pre-Filtering (PFMPF) and the Pixel-Based Hierarchical-Feature Adaboosting (PBHFA) method to effectively solve the above-mentioned problems in cascade Adaboosting. The two stages both provide far less training time than that of the cascade Adaboosting and thus reduce the computation complexity in face-detection tasks. In particular, the proposed PFMPF can effectively filter out more than 85% nonface in an image and the remaining few face candidates are then secondly filtered with a single PBHF Adaboost strong classifier. Given a <formula formulatype="inline" xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex Notation="TeX">${\rm M}\times{\rm N}$</tex></formula> sub-window, the number of possible PBH features is simplified down to a level less than <formula formulatype="inline" xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"><tex Notation="TeX">${\rm M}\times{\rm N}$</tex></formula> , which significantly reduces the length of the training period by a factor of 1500. Moreover, when the two-stage hybrid face detection scheme are employed for practical face-detection tasks, the complexity is still lower than that of the integral-image based approach in the traditional Adaboosting method. Experimental results obtained using the gray feret database show that the proposed two-stage hybrid face detection scheme is significantly more effective than Haar-like features.

A New Skin and Frontal Face Detector by selecting Optimized Colour Spaces

In this paper a new method to detect faces in colour images, is proposed which has wide range of application in HumanComputer interaction, face processing and illness diagnosis (specially mental disease). Skin colour, lip position and face shape information are the key parameters for developing a fussy rule based classifier to extract face candidate from an image. In first step, skin regions are identified in HSI colour space, using fuzzy system, applying distance of each pixel colour to skin colour cluster as input and produce a skin-likelihood image in output. Next step is searching lip area in the region owning the most likelihood of belonging to skin. This stage is implemented in normalized RGB colour space by fuzzy system its input is r-g value. To extract face regions, fuzzy rule based system is used, applying face and lip position, lip area data and face shape to decide which skin region is human face. With this algorithm, 98.75%, 86.6% detection rates are achieved, respectively for frontal and near frontal. Revising system for detecting profile and rotating faces associated with exactly determine lip area, is the next goal.

A real-time personal authentication system based on incremental feature extraction and classification of audiovisual information

We propose a new approach to a real-time personal authentication system based on incrementally updated visual (face) and audio (voice) features of persons. The proposed system consists of real-time face detection, incremental audiovisual feature extraction, and incremental neural classifier model with long-term memory. The face detection part, a biologically motivated face-color preferable selective attention model first localizes face candidate regions in natural scenes, and then the Adaboost-based face detection identifies human faces from the localized face-candidate regions. The mel-frequency cepstral coefficient is used for vocal feature extraction of speakers. Moreover, incremental principal component analysis (IPCA) is used to reduce the dimensions of audiovisual features and to update them incrementally. The features extracted by IPCA is fed to the resource allocating network with long-term memory which learns facial and vocal features incrementally and recognizes faces in real time. Experimental results show that the proposed system can enhance the test performance incrementally without serious forgetting. In addition, a multi-modal (facial and vocal) feature effectively increases the robustness of the personal authentication system in noisy environments.

Neural Illustration based Creature Recognition System

Face detection is one of the challenging problems in image processing. In this paper we have proposed a new technology named neural GUI based human recognition robotics. This system contains the skin color which is the main feature of faces for detection, and then the skin face candidate is examined by using the neural GUI, which learn from the feature of faces to classify whether the original image includes a face or not. Finally the classification is based on GUI approach. The experiment results on upright frontal color face images from the internet show an excellent detection rate.

A new method for face detection in colour images for emotional bio-robots

Emotional bio-robots have become a hot research topic in last two decades. Though there have been some progress in research, design and development of various emotional bio-robots, few of them can be used in practical applications. The study of emotional bio-robots demands multi-disciplinary co-operation. It involves computer science, artificial intelligence, 3D computation, engineering system modelling, analysis and simulation, bionics engineering, automatic control, image processing and pattern recognition etc. Among them, face detection belongs to image processing and pattern recognition. An emotional robot must have the ability to recognize various objects, particularly, it is very important for a bio-robot to be able to recognize human faces from an image. In this paper, a face detection method is proposed for identifying any human faces in colour images using human skin model and eye detection method. Firstly, this method can be used to detect skin regions from the input colour image after normalizing its luminance. Then, all face candidates are identified using an eye detection method. Comparing with existing algorithms, this method only relies on the colour and geometrical data of human face rather than using training datasets. From experimental results, it is shown that this method is effective and fast and it can be applied to the development of an emotional bio-robot with further improvements of its speed and accuracy.

Person re-identification in TV series using robust face recognition and user feedback

In this paper, we present a system for person re-identification in TV series. In the context of video retrieval, person re-identification refers to the task where a user clicks on a person in a video frame and the system then finds other occurrences of the same person in the same or different videos. The main characteristic of this scenario is that no previously collected training data is available, so no person-specific models can be trained in advance. Additionally, the query data is limited to the image that the user clicks on. These conditions pose a great challenge to the re-identification system, which has to find the same person in other shots despite large variations in the person's appearance. In the study, facial appearance is used as the re-identification cue, since, in contrast to surveillance-oriented re-identification studies, the person can have different clothing in different shots. In order to increase the amount of available face data, the proposed system employs a face tracker that can track faces up to full profile views. This makes it possible to use a profile face image as query image and also to retrieve images with non-frontal poses. It also provides temporal association of the face images in the video, so that instead of using single images for query or target, whole tracks can be used. A fast and robust face recognition algorithm is used to find matching faces. If the match result is highly confident, our system adds the matching face track to the query set. Finally, if the user is not satisfied with the number of returned results, the system can present a small number of candidate face images and lets the user confirm the ones that belong to the queried person. These features help to increase the variation in the query set, making it possible to retrieve results with different poses, illumination conditions, etc. The system is extensively evaluated on two episodes of the TV series Coupling, showing very promising results.

WAITING TIME AS A PRICE FOR DECEASED DONOR KIDNEYS*

I. INTRODUCTION Kidneys for transplantation are supplied by deceased and living organ donors. Candidates for kidney transplantation face the choice of joining the waiting list for deceased donor kidneys, where many wait over 3 years before receiving an organ, or finding a living donor. Patients with living donors are transplanted immediately. In this study, I estimate the impact of waiting time on the demand for deceased versus living donor kidneys. I hypothesize that waiting time acts like a price for deceased donor organs. As waiting time increases, new transplant candidates will substitute away from the deceased donor waiting list toward living donor transplantation. This study follows extensive work on the impact of waiting times for surgery and other procedures in the British National Health Service (see, e.g., Cullis, Jones, and Propper 2000; Martin et al. 2007; Windmeijer, Gravelle, and Hoonhout 2005). In this setting, private treatment or no care form the outside option. A difficulty faced by researchers in estimating the impact of waiting time on demand is reverse causation: there is a mechanistic pathway leading from demand to waiting time and a behavioral pathway leading from waiting time to demand. I use a structural model of the wailing list to construct a predicted waiting time measure that isolates variation in waiting time due to regional differences in the demand for and supply of deceased donor kidneys. I find that use of living donor transplantation is moderately responsive to the supply of deceased donors. II. BACKGROUND Patients with end stage renal disease who are suitable candidates for kidney transplantation are referred to transplant centers for evaluation. During the initial appointment, patients are encouraged to consider living donor transplantation and to solicit possible donors from among family and friends. Living donor transplantation offers a number of advantages. There is no waiting list, and living donor transplant recipients experience better posttransplant outcomes. The only disadvantage is that it subjects healthy living donors to a medical procedure from which they derive no direct benefit, in apparent violation of the first, do no harm precept of medical ethics. Although the short-term mortality risks are extremely small (Ibrahim et al. 2009; Ingelfinger 2005; Segev et al. 2010), all donors suffer from the discomfort associated with undergoing an invasive operation. Potential living donors are subjected to an extensive physical and psychological evaluation. Traditionally, donors who were not immunologically compatible with the recipient were ruled out. Now it is increasingly common to perform transplants where donor and recipient do not match on any of the six human leukocyte antigens used to assess compatibility. This trend reflects improvements in immunosuppressive drugs and possibly, consistent with the study hypothesis, longer wait times for deceased donor kidneys. (1), (2) Patients unable to find a living donor must wait for a kidney from a deceased donor. (3) Deceased donor kidneys are allocated to patients on the waiting list based on organ procurement organization (OPO) region, waiting time, and immunologic compatibility. (4) OPOs are the 58 regional entities that oversee the consent and organ removal process. A kidney from a deceased donor is first offered to wait-listed patients in the same OPO region. If the kidney is not used by any of the local transplant centers, then it is offered to patients in other regions. (5) In 2005, 60% of kidneys were transplanted locally. The remainder were allocated to patients outside the local region based on accumulated waiting time and compatibility. There are a number of harms and risks associated with waiting for a deceased donor transplant. The longer the wait, the greater is the likelihood the patient will die before receiving a transplant. Longer waiting times are associated with worse posttransplant outcomes (Meier-Kriesche et al. …

A novel fuzzy rule base system for pose independent faces detection

Human face detection plays an important role in a wide range of applications such as face recognition, surveillance systems, video tracking applications, and image database management. In this paper, a novel fuzzy rule-based system for pose, size, and position independent face detection in color images is proposed. Subtractive clustering method is also applied to decide on the numbers of membership functions. In the proposed system, skin-color, lips position, face shape information and ear texture properties are the key parameters fed to the fuzzy rule-based classifier to extract face candidate in an image. Furthermore, the applied threshold on the face candidates is optimized by genetic algorithm. The proposed system consists of two main stages: the frontal/near frontal face detections and the profile face detection. In the first stage, skin and lips regions are identified in HSI color space, using fuzzy schemes, where the distances of each pixel color to skin-color and lips-color clusters are applied as the input and skin-likelihood and lips-like images are produced as the output. Then, the labeled skin and lips regions are presented to both frontal and profile face detection algorithms. A fuzzy rule-based containing the face and lips position data, along with the lips area and face shape are employed to extract the frontal/near frontal face regions. On the other hand, the profile face detection algorithm uses a geometric moments-based ear texture classification to verify its outcomes. The proposed method is tried on various databases, including HHI, Champion, Caltech, Bao, Essex and IMM databases. It shows about 98, 96 and 90% correct detection rates over 783 samples, in frontal, near frontal and profile face images, respectively.

웃음 치료 훈련을 위한 웃음 표정 인식 시스템 개발

본 논문은 실시간 카메라 영상으로부터 얼굴을 검출하고 얼굴 표정을 인식하여 웃음 치료훈련을 할 수 있는 시스템을 제안한다. 제안된 시스템은 카메라 영상으로부터 Haar-like 특징을 이용하여 얼굴 후보 영역을 검출한 다음, SVM분류기를 이용하여 얼굴 후보 영역이 얼굴 영상인지 아닌지를 검증한다. 그 다음에는 검출된 얼굴 영상에 대해, 조명의 영향을 최소화하기 위한 방법으로 히스토그램 매칭을 이용한 조명 정규화를 수행한다. 표정 인식 단계에서는 PCA를 사용하여 얼굴 특징 벡터를 획득한 후 다층퍼셉트론 인공신경망을 이용해 실시간으로 웃음표정을 인식하였다. 본 논문에서 개발된 시스템은 실시간으로 사용자의 웃음 표정을 인식하여 웃음 양을 화면에 표시해 줌으로써 사용자 스스로 웃음 훈련을 할 수 있게 해 준다. 실험 결과에 따르면, 본 논문에서 제안한 방법은 SVM 분류기를 통한 얼굴 후보 영역 검증과 히스토그램 매칭을 이용한 조명정규화를 이용하여 웃음 표정 인식률을 향상시켰다. In this paper, we proposed a recognition system of smile facial expression for smile treatment training. The proposed system detects face candidate regions by using Haar-like features from camera images. After that, it verifies if the detected face candidate region is a face or non-face by using SVM(Support Vector Machine) classification. For the detected face image, it applies illumination normalization based on histogram matching in order to minimize the effect of illumination change. In the facial expression recognition step, it computes facial feature vector by using PCA(Principal Component Analysis) and recognizes smile expression by using a multilayer perceptron artificial network. The proposed system let the user train smile expression by recognizing the user's smile expression in real-time and displaying the amount of smile expression. Experimental result show that the proposed system improve the correct recognition rate by using face region verification based on SVM and using illumination normalization based on histogram matching.

Multi-pose face detection based on facial features and AdaBoost algorithm

An improved multi-pose face detection algorithm based on facial features and AdaBoost algorithm was proposed.Making full use of facial skin color information firstly,most of the background regions could be quickly excluded.After detecting eyes and mouth,the approximate frontal face candidate regions were segmented according to face orientation decided by the geometric features of the eyes and mouth regions.At last,the face candidate regions were classified by AdaBoost algorithm.The experimental results demonstrate that the algorithm can further improve the multi-pose face detection accuracy and is highly robust to facial expression and occlusion.

Fuzzy system-based real-time face tracking in a multi-subject environment with a pan-tilt-zoom camera

This paper proposes real-time face tracking in a multi-subject environment with a pan-tilt-zoom camera using the fuzzy system technique. Tracking is based on detected faces in the Hue–Saturation–Value (HSV) color space. To detect faces, a fuzzy classifier segments skin colors in the HS color space. To reduce the influence of illumination, a fuzzy system is designed to adaptively determine the fuzzy classifier segmentation threshold according the V color space of an image. Detected skin regions are considered face candidates. Shape and color features serve as another fuzzy classifier inputs, leading to a final detection. For face tracking, a Kalman filter algorithm predicts face detection regions and corrects tracking trajectory. When no face is detected, the consecutive frame difference is employed to track the moving person to avoid tracking lost. To track a specific person among multiple persons, a clothing color histogram in the HS space serves as the determination criterion. The performance of the proposed face detection method is compared to other real-time face detection methods. In real-time operations, the tracking system uses camera panning, tilting, and zooming operations to keep the tracked person within the camera view and maintain a suitable face size in the image.

Interesting faces: A graph-based approach for finding people in news

In this study, we propose a method for finding people in large news photograph and video collections. Our method exploits the multi-modal nature of these data sets to recognize people and does not require any supervisory input. It first uses the name of the person to populate an initial set of candidate faces. From this set, which is likely to include the faces of other people, it selects the group of most similar faces corresponding to the queried person in a variety of conditions. Our main contribution is to transform the problem of recognizing the faces of the queried person in a set of candidate faces to the problem of finding the highly connected sub-graph (the densest component) in a graph representing the similarities of faces. We also propose a novel technique for finding the similarities of faces by matching interest points extracted from the faces. The proposed method further allows the classification of new faces without needing to re-build the graph. The experiments are performed on two data sets: thousands of news photographs from Yahoo! news and over 200 news videos from TRECVid2004. The results show that the proposed method provides significant improvements over text-based methods.

A robust facial feature detection on mobile robot platform

Human face analysis on the mobile robot vision system should cope with difficult problems such as face pose variations, illumination changes, and complex backgrounds, in which problems are mainly induced from the movement of its platform. In this paper, in order to overcome such problems, an efficient facial feature detection approach based on local image region and direct pixel-intensity distributions is presented. We propose two novel concepts; the directional template for evaluating intensity distributions and the edge-like blob map image with multiple strength intensity. Using this blob map image, we show that the locations of major facial features—two eyes and a mouth—can be reliably estimated. Without the boundary information of facial area, final candidate face region is determined by both obtained locations of facial features and weighted correlations with stored facial templates.

Gappy PCA Classification for Occlusion Tolerant 3D Face Detection

This paper presents an innovative approach for the detection of faces in three dimensional scenes. The method is tolerant against partial occlusions produced by the presence of any kind of object. The detection algorithm uses invariant properties of the surfaces to segment salient facial features, namely the eyes and the nose. At least two facial features must be clearly visible in order to perform face detection. Candidate faces are then registered using an ICP (Iterative Correspondent Point) based approach aimed to avoid those samples which belong to the occluding objects. The final face versus non-face discrimination is computed by a Gappy PCA (GPCA) classifier which is able to classify candidate faces using only those regions of the surface which are considered to be non-occluded. The algorithm has been tested using the UND database obtaining 100% of correct detection and only one false alarm. The database has been then processed with an artificial occlusions generator producing realistic acquisitions that emulate unconstrained scenarios. A rate of 89.8% of correct detections shows that 3D data is particularly suited for handling occluding objects. The results have been also verified on a small test set containing real world occlusions obtaining 90.4% of correctly detected faces. The proposed approach can be used to improve the robustness of all those systems requiring a face detection stage in non-controlled scenarios.

Novel face-detection method under various environments

We propose a method to detect a face with different poses under various environments. On the basis of skin color information, skin regions are first extracted from an input image. Next, the shoulder part is cut out by using shape information and the head part is then identified as a face candidate. For a face candidate, a set of geometric features is applied to determine if it is a profile face. If not, then a set of eyelike rectangles extracted from the face candidate and the lighting distribution are used to determine if the face candidate is a nonprofile face. Experimental results show that the proposed method is robust under a wide range of lighting conditions, different poses, and races. The detection rate for the HHI face database is 93.68%. For the Champion face database, the detection rate is 95.15%.

Haar-like 특징과 템플릿을 이용한 귀 검출

Ear detection in an image processing is the one of the important area in biometrics. In this paper we propose a human ear detection algorithm with side face images. First, we search a face candidate area in an input image by using skin-color model and try to find an ear area based on Haar-like feature. Then, to verity whether it is the ear area or not, we use the template which is excellent object classification compare to recognize the characters in the plate. In this experiment, the proposed method showed that the processing speed is improved by 60% than previous works and the detection success rate is 92%.

Face swapping

In this paper, we present a complete system for automatic face replacement in images. Our system uses a large library of face images created automatically by downloading images from the internet, extracting faces using face detection software, and aligning each extracted face to a common coordinate system. This library is constructed off-line, once, and can be efficiently accessed during face replacement. Our replacement algorithm has three main stages. First, given an input image, we detect all faces that are present, align them to the coordinate system used by our face library, and select candidate face images from our face library that are similar to the input face in appearance and pose. Second, we adjust the pose, lighting, and color of the candidate face images to match the appearance of those in the input image, and seamlessly blend in the results. Third, we rank the blended candidate replacements by computing a match distance over the overlap region. Our approach requires no 3D model, is fully automatic, and generates highly plausible results across a wide range of skin tones, lighting conditions, and viewpoints. We show how our approach can be used for a variety of applications including face de-identification and the creation of appealing group photographs from a set of images. We conclude with a user study that validates the high quality of our replacement results, and a discussion on the current limitations of our system.

Exploiting Voronoi diagram properties in face segmentation and feature extraction

Segmentation of human faces from still images is a research field of rapidly increasing interest. Although the field encounters several challenges, this paper seeks to present a novel face segmentation and facial feature extraction algorithm for gray intensity images (each containing a single face object). Face location and extraction must first be performed to obtain the approximate, if not exact, representation of a given face in an image. The proposed approach is based on the Voronoi diagram (VD), a well-known technique in computational geometry, which generates clusters of intensity values using information from the vertices of the external boundary of Delaunay triangulation ( DT). In this way, it is possible to produce segmented image regions. A greedy search algorithm looks for a particular face candidate by focusing its action in elliptical-like regions. VD is presently employed in many fields, but researchers primarily focus on its use in skeletonization and for generating Euclidean distances; this work exploits the triangulations (i.e., Delaunay) generated by the VD for use in this field. A distance transformation is applied to segment face features. We used the BioID face database to test our algorithm. We obtained promising results: 95.14% of faces were correctly segmented; 90.2% of eyes were detected and a 98.03% detection rate was obtained for mouth and nose.