scispace - formally typeset
Search or ask a question

Showing papers on "Three-dimensional face recognition published in 1999"



Journal ArticleDOI
TL;DR: Name-It, a system that associates faces and names in news videos, takes a multimodal video analysis approach: face sequence extraction and similarity evaluation from videos, name extraction from transcripts, and video-caption recognition.
Abstract: We developed Name-It, a system that associates faces and names in news videos. It processes information from the videos and can infer possible name candidates for a given face or locate a face in news videos by name. To accomplish this task, the system takes a multimodal video analysis approach: face sequence extraction and similarity evaluation from videos, name extraction from transcripts, and video-caption recognition.

311 citations


Proceedings ArticleDOI
24 Oct 1999
TL;DR: The results show that the use of the color information embedded in a eigen approach, improve the recognition rate when compared to the same scheme which uses only the luminance information.
Abstract: A common feature found in practically all technical approaches proposed for face recognition is the use of only the luminance information associated to the face image. One may wonder if this is due to the low importance of the color information in face recognition or due to other less technical reasons such as the no availability of color image database. Motivated by this reasoning, we have performed a variety of tests using a global eigen approach developed previously, which has been modified to cope with the color information. Our results show that the use of the color information embedded in a eigen approach, improve the recognition rate when compared to the same scheme which uses only the luminance information.

192 citations


Journal ArticleDOI
TL;DR: These experiments indicate that significant gains are to be realized in both speed and recognition accuracy by using a contour representation in handwriting applications.
Abstract: Contour representations of binary images of handwritten words afford considerable reduction in storage requirements while providing lossless representation. On the other hand, the one-dimensional nature of contours presents interesting challenges for processing images for handwritten word recognition. Our experiments indicate that significant gains are to be realized in both speed and recognition accuracy by using a contour representation in handwriting applications.

87 citations


Proceedings ArticleDOI
26 Sep 1999
TL;DR: This work extends SVMs to model the 2D appearance of human faces which undergo nonlinear change across the view sphere and enables simultaneous multi-view face detection and pose estimation at near-frame rate.
Abstract: Support vector machines have shown great potential for learning classification functions that can be applied to object recognition. In this work, we extend SVMs to model the 2D appearance of human faces which undergo nonlinear change across the view sphere. The model enables simultaneous multi-view face detection and pose estimation at near-frame rate.

83 citations


Journal ArticleDOI
TL;DR: A novel image-based face recognition algorithm that uses a set of random rectilinear line segments of 2D face image views as the underlying image representation, together with a nearest-neighbor classifier as the line matching scheme, which achieves high generalization recognition rates for rotations both in and out of the plane.
Abstract: Much research in human face recognition involves fronto-parallel face images, constrained rotations in and out of the plane, and operates under strict imaging conditions such as controlled illumination and limited facial expressions. Face recognition using multiple views in the viewing sphere is a more difficult task since face rotations out of the imaging plane can introduce occlusion of facial structures. In this paper, we propose a novel image-based face recognition algorithm that uses a set of random rectilinear line segments of 2D face image views as the underlying image representation, together with a nearest-neighbor classifier as the line matching scheme. The combination of 1D line segments exploits the inherent coherence in one or more 2D face image views in the viewing sphere. The algorithm achieves high generalization recognition rates for rotations both in and out of the plane, is robust to scaling, and is computationally efficient. Results show that the classification accuracy of the algorithm is superior compared with benchmark algorithms and is able to recognize test views in quasi-real-time.

75 citations


Proceedings ArticleDOI
23 Jun 1999
TL;DR: It is demonstrated that face recognition can be considerably improved by the analysis of video sequences, and the method presented is widely applicable in many multi-class interpretation problems.
Abstract: We present a quantitative evaluation of an algorithm for model-based face recognition. The algorithm actively learns how individual faces vary through video sequences, providing on-line suppression of confounding factors such as expression, lighting and pose. By actively decoupling sources of image variation, the algorithm provides a framework in which identity evidence can be integrated over a sequence. We demonstrate that face recognition can be considerably improved by the analysis of video sequences. The method presented is widely applicable in many multi-class interpretation problems.

63 citations


Proceedings ArticleDOI
24 Oct 1999
TL;DR: This paper presents an advanced face recognition system that is based on the use of Pseudo 2-D HMMs and coefficients of the2-D DCT as features that works directly with JPEG-compressed face images, without any necessity of completely decompressing the image before recognition.
Abstract: This paper presents an advanced face recognition system that is based on the use of Pseudo 2-D HMMs and coefficients of the 2-D DCT as features. A major advantage of our approach is the fact that our face recognition system works directly with JPEG-compressed face images, i.e. it uses directly the DCT-features provided by the JPEG standard, without any necessity of completely decompressing the image before recognition. The recognition rates on the Olivetti Research Laboratory (ORL) face database are 100% for the original images and 99.5% for JPEG compressed domain recognition. A comparison with other face recognition systems evaluated on the ORL database, shows that these are the best recognition results on this database.

52 citations


Proceedings ArticleDOI
01 Jun 1999
TL;DR: A Bayesian recognition framework in which a model of the whole face is enhanced by models of facial feature position and appearances, and facial appearance matching is improved by facial expression matching.
Abstract: We present a Bayesian recognition framework in which a model of the whole face is enhanced by models of facial feature position and appearances. Face recognition and facial expression recognition are carried out using maximum likelihood decisions. The algorithm finds the model and facial expression that maximizes the likelihood of a test image. In this framework, facial appearance matching is improved by facial expression matching. Also, changes in facial features due to expressions are used together with facial deformation. Patterns to jointly perform expression recognition. In our current implementation, the face is divided into 9 facial features grouped in 4 regions which are detected and tracked automatically in video segments. The feature images are modeled using Gaussian distributions on a principal component sub-space. The training procedure is supervised; we use video segments of people in which the facial expressions have been segmented and labeled by hand. We report results on face and facial expression recognition using a video database of 18 people and 6 expressions.

49 citations


Proceedings ArticleDOI
15 Sep 1999
TL;DR: A computer system that can locate and track a subject's head in a complex background and then recognize the person by comparing characteristics of the face to those of known individuals is developed.
Abstract: In this paper we developed a computer system that can locate and track a subject's head in a complex background and then recognize the person by comparing characteristics of the face to those of known individuals. The computational approach taken in this system is motivated by color and motion Information and PCA (principal component analysis). Our approach treats the face recognition problem as a two-dimensional (2-D) problem rather than three-dimensional geometry. So, the problem is easier to treat. The system functions by two steps, first, extracting face image in a complex background using difference image and color model, and second, projecting pre-extracted face images onto a feature space that represents the significant variations among known face images. We use this weight vector to recognize each individual. Several evaluation methods of this weight vector are attempted in this paper.

47 citations


Journal ArticleDOI
TL;DR: The article presents the latest trends in computer-based human face recognition by introducing methods of facial pattern representation intended to identify persons using frontal face images as well as recent studies in extending those methods by giving flexibility in terms of face orientation and view.
Abstract: The article presents the latest trends in computer-based human face recognition. First, methods of facial pattern representation intended to identify persons using frontal face images are introduced as well as recent studies in extending those methods by giving flexibility in terms of face orientation and view. The FERET face recognition project, which is one of the driving forces to promote robustness in face recognition technology, is considered. In addition, computer face recognition is treated as an element of media processing for content search and editing of visual databases, and studies in computer face recognition that model human cognition in the face recognition process are surveyed. Finally, the state of the art in extracting facial patterns from visual scenes is analyzed. © 1999 Scripta Technica, Syst Comp Jpn, 30(10): 76–89, 1999

Proceedings ArticleDOI
22 Jun 1999
TL;DR: A new face recognition system that can be used to index (and thus retrieve) images and videos of a database of faces, based on identifying frontal faces, that allows reasonable variability in facial expressions, illumination conditions, and occlusions caused by eye-wear or items of clothing such as scarves.
Abstract: The paper introduces a new face recognition system that can be used to index (and thus retrieve) images and videos of a database of faces. New face recognition approaches are needed because, although much progress has been made to identify face taken from different viewpoints, we still cannot robustly identify faces under different illumination conditions, or when the facial expression changes, or when a part of the face is occluded on account of glasses or parts of clothing. When face recognition methods have worked in the past, it was only when all possible "image variations" were learned. Principal components analysis (PCA) and Fisher Discriminant Analysis (FDA) are well-known cases of such methods. We present a different approach to the indexing of face images. Our approach is based on identifying frontal faces and it allows reasonable variability in facial expressions, illumination conditions, and occlusions caused by eye-wear or items of clothing such as scarves. We divide a face image into n different regions, analyze each region with PCA, and then use a Bayesian approach to find the best possible global match between a query image and a database image. The relationships between the n parts is modeled by using Hidden Markov Models (HMMs).

Proceedings ArticleDOI
A.Z. Kouzani1
01 Jan 1999
TL;DR: Based on the concepts of linear object classes and the principal components analysis, an illumination-effects compensation method is presented to transform an arbitrary- lit face image whose illumination effects are pre-determined, into a front-lit face image.
Abstract: Based on the concepts of linear object classes and the principal components analysis, an illumination-effects compensation method is presented to transform an arbitrary-lit face image whose illumination effects are pre-determined, into a front-lit face image.

Proceedings ArticleDOI
23 Feb 1999
TL;DR: A simple technique for identification of human faces in cluttered scenes based on neural nets based on Fourier descriptors, which results in reduction of computational complexity and thus decreasing the time and memory needed during the testing of an image.
Abstract: Automatic recognition of human faces is a significant problem in the development and application of pattern recognition. We introduce a simple technique for identification of human faces in cluttered scenes based on neural nets. In the detection phase, neural nets are used to test whether a window of 20/spl times/20 pixels contains a face or not. A major difficulty in the learning process comes from the large database required for face/nonface images. We solve this problem by dividing these data into two groups. Such a division results in reduction of computational complexity and thus decreasing the time and memory needed during the testing of an image. For the recognition phase, feature measurements are made through Fourier descriptors. Such features are used as input to the neural classifier for training and recognition of ten human faces. Simulation results for the proposed algorithm show a good performance during testing.

Proceedings ArticleDOI
Andrew W. Senior1
26 Sep 1999
TL;DR: The application of a face recognition system to video indexing, with the joint purpose of labelling faces in the video, and identifying speakers, is described and a new method of aggregating multiple Gabor jet representations for a whole sequence is described.
Abstract: Face recognition has recently attracted increasing attention and is beginning to be applied in a variety of domains, predominantly for security, but also for video indexing. This paper describes the application of a face recognition system to video indexing, with the joint purpose of labelling faces in the video, and identifying speakers. The face recognition system can be used to supplement acoustic speaker identification, when the speaker's face is shown, to allow indexing of the speakers, as well as the selection of the correct speaker-dependent model for speech transcription. This paper describes the feature detection and recognition methods used by the system, and describes a new method of aggregating multiple Gabor jet representations for a whole sequence. Several approaches to using such aggregate representation for recognition of faces in image sequences are compared. Results are presented showing a significant improvement in recognition rates when the the whole sequence is used instead of a single image of the face.

Proceedings ArticleDOI
24 Oct 1999
TL;DR: First results for face recognition of video sequences are presented and the main and final objective is to develop a tool to be used in the MPEG-7 standardization effort to help video indexing activities.
Abstract: An integral scheme that provides a global eigen approach to the problem of face recognition of still images has been presented by Lorente and Torres, (1998). The scheme is based on the representation of the face images using the so called eigenfaces, generated performing a PCA (Principal Components Analysis). The data base used was designed for still image recognition and the corresponding images were very controlled. That is, the test images had controlled expression, orientation and lighting variations. Preliminary results were shown using only a frontal view image by person in the training set. In this paper, we present our first results for face recognition of video sequences. To that end, we have modified our original scheme in such a way that is able to cope with the different face conditions present in a video sequence. The main and final objective is to develop a tool to be used in the MPEG-7 standardization effort to help video indexing activities. The system is not yet fully automatic, but an automatic facial point location is under development. Good results have been obtained using the video test sequences used in the MPEG-7 evaluation group.

Proceedings ArticleDOI
01 Jan 1999
TL;DR: In this article, the authors evaluated lip features for person recognition and compared the performance with that of the acoustic signal, and found that the performance of the upper lip was considerably better than the lower lip, achieving 15% and 35% identification error rates respectively.
Abstract: This paper evaluates lip features for person recognition, and compares the performance with that of the acoustic signal. Recognition accuracy is found to be equivalent in the two domains, agreeing with the findings of Chibelushi (1997). The optimum dynamic window length for both acoustic and visual modalities is found to be about 100 ms. Recognition performance of the upper lip is considerably better than the lower lip, achieving 15% and 35% identification error rates respectively, using a single digit test and training token.

Journal Article
TL;DR: A fast face detection algorithm based on regional feature is proposed that can find the face rapidly with relative high accuracy, with little limit to the complexity of background, lighting condition, face size, person number, resolution of the image etc.
Abstract: Automatic face recognition (AFR) is one of the most attractive and challenging tasks in fields of computer vision and pattern recognition, the first critical step of AFR is face detection. A fast face detection algorithm based on regional feature is proposed. In contrast with the award winning mosaic method, three main improvements were introduced here: 1). organ based blocking scheme and more intuitive mosaic rule design, 2). sub block type adaptive technique according to different face shape, 3). fast organ based rough detection and hierarchical local searching method. Also directional gradient statistics were used as sub block feature instead of absolute gray value ones. Experimental results show that the algorithm can find the face rapidly with relative high accuracy, with little limit to the complexity of background, lighting condition, face size, person number, resolution of the image etc.

Proceedings ArticleDOI
05 Oct 1999
TL;DR: A novel face identification method that combines the eigenfaces theory with the Neural Nets, and a neural net classifier that performs the identification process is presented.
Abstract: This paper describes a novel face identification method that combines the eigenfaces theory with the Neural Nets. We use the eigenfaces methodology in order to reduce the dimensionality of the input image, and a neural net classifier that performs the identification process. The method presented recognizes faces in the presence of variations in facial expression, facial details and lighting conditions. A recognition rate of more than 87% has been achieved, while the classical method of Turk and Pentland achieves a 75.5%.

Proceedings ArticleDOI
01 Jan 1999
TL;DR: The sequence of poses can be reduced into a trajectory on the two-dimensional eigenspace with preserving the main features in gesture, so that the gesture recognition equals the character recognition.
Abstract: This paper describes a novel method for gesture recognition using character recognition techniques on two-dimensional eigenspace. An image-based approach can capture human body poses in 3D motion from multiple image sequences. The sequence of poses can be reduced into a trajectory on the two-dimensional eigenspace with preserving the main features in gesture, so that the gesture recognition equals the character recognition. Experiments for the gesture recognition using some character recognition techniques show our method is useful.

Proceedings ArticleDOI
01 Jan 1999
TL;DR: This paper describes a system for recognizing human faces from a stereo image pair out of a large database with one image pair per person, stored as image meshes which contain the position and disparity for 40 feature points extracted from the image pair.
Abstract: This paper describes a system for recognizing human faces from a stereo image pair out of a large database with one image pair per person. The faces are stored as image meshes which contain the position and disparity for 40 feature points extracted from the image pair. Texture information, described by the outputs of a set of Gabor wavelets is stored for each feature point. The Gabor wavelets are also used in the disparity estimation process to resolve a 3-D representation of the face. The recognition is conducted by examining how well the stored meshes can be projected onto the image pair to be tested, using camera calibration data. This is important as recognition is independent of pose, camera geometry, distance from camera and lighting conditions.

Proceedings ArticleDOI
05 Sep 1999
TL;DR: Extended experiments carried out in a test-bed of 6406 face images, have shown that the face detection accuracy is increased significantly when non-linear and probabilistic illumination equalizers pre-process the sub-images.
Abstract: In this paper we present a neural detector of frontal faces in gray scale images under arbitrary face size, orientation, facial expression, skin color, lighting conditions and background environment. In a two-level process, a window normalization module reduces the variability of the features and a neural classifier generates multiple face position hypotheses. Extended experiments carried out in a test-bed of 6406 face images, have shown that the face detection accuracy is increased significantly when non-linear and probabilistic illumination equalizers pre-process the sub-images. Moreover, better results can be achieved in case of training the neural detector using positional and orientation normalized face examples. In this case the neural face detector has the capability to locate both position and orientation of a face. In the multiple face position hypotheses generated by the proposed neural method, 98.3% detection accuracy, the highest reported in the literature, was measured.

Proceedings ArticleDOI
01 Jan 1999
TL;DR: A person verification system based on facial profile views and features extracted from speech whose outputs are fused after a normalization step is described, which shows that integration of the face profile and speech information results in superior performance to that of its subsystems.
Abstract: This paper describes a person verification system based on facial profile views and features extracted from speech. The system is comprised of two non-homogeneous classifiers whose outputs are fused after a normalization step. Experiments are reported which show that integration of the face profile and speech information results in superior performance to that of its subsystems. Additionally, the performance of the combined system in noisy conditions is shown to be more robust than the speech-based subsystem alone.

Proceedings ArticleDOI
27 Sep 1999
TL;DR: The paper describes a method for face recognition based on the eigenimage technique that allows images to be represented by a limited set of parameters and be compared according to simple similarity criteria for classification or retrieval purposes.
Abstract: We describe a method for face recognition based on the eigenimage technique that allows images to be represented by a limited set of parameters and be compared according to simple similarity criteria for classification or retrieval purposes. Our method is not applied to the whole image of the face, but to sub-images representing the most salient face components (eyes, nose, mouth). The method is able to recognise with good precision faces having different head postures on the image plane, because faces are straightened through a rotation transform. Moreover the paper reports results we have achieved with such a method on two publicly-available reference sets of images.

Proceedings ArticleDOI
01 Dec 1999
TL;DR: The complex facial pattern matching process is reduced to an elastic graph system matching of facial contours by simplifying the traditional dynamic link model and integrates it with the active contour model for feature extraction.
Abstract: Face recognition relies heavily on feature extraction and the classification of features in the process of pattern recognition. Existing methods tend to address the problem with some tradeoff between the speed and accuracy in the process. In this paper, a system known as elastic graph dynamic link model (EGDLM) is proposed to provide an effective and reliable solution. The model simplifies the traditional dynamic link model and integrates it with the active contour model for feature extraction. The complex facial pattern matching process is reduced to an elastic graph system matching of facial contours. A database of 1020 facial images was used for model testing and experimental results indicate an improvement of average recognition speed by more than 1000 times, and an overall recognition rate of over 85%.

Proceedings ArticleDOI
04 Oct 1999
TL;DR: A fast 3D face reconstruction algorithm using isoluminance lines from stereo images, which is effective even when only a few edges can be detected, and very efficient algorithm do search for irises in 3D space, which can then be used to identify the face position and direction.
Abstract: A segment-based stereo vision system has been modified to reconstruct 3D facial models. Because usually it is difficult to get enough edges to reconstruct a 3D facial model with edge-based stereo algorithms, most research on 3D face model is based on correlation stereo algorithms, even though these algorithms are often time-consuming. We present a fast 3D face reconstruction algorithm using isoluminance lines from stereo images, which is effective even when only a few edges can be detected. We also introduced very efficient algorithm do search for irises in 3D space, which can then be used to identify the face position and direction. Our experiments show that our 3D facial model is suitable for 3D face recognition.

Proceedings ArticleDOI
27 Sep 1999
TL;DR: An algorithm is presented which constitutes the fine analysis part of a multiscale face detection algorithm in a complex background and which detects, with a subpixellic precision, the center and the ray of the eyeballs of a person's eyes, and which achieves a geometrical normalization of the face's image under scale, rotation and translation.
Abstract: In a face recognition system, the normalization of the faces under scale, rotation and translation, as well as under contrast and brightness variation, is a crucial step for the extraction of stable features used to recognize a face. A way to improve the speed and the recognition performance of the overall system, is to have an accurate normalization of the images, especially in scale. Here, we present an algorithm, which constitutes the fine analysis part of a multiscale face detection algorithm in a complex background, and which detects, with a subpixellic precision, the center and the ray of the eyeballs of a person's eyes. The segment joining these two points is then used to achieve a geometrical normalization of the face's image under scale, rotation and translation. One of the major advantages of this method is to reduce greatly the number of possible scales used during the face recognition process.

Proceedings ArticleDOI
10 Jul 1999
TL;DR: A new technique is discussed to recognize human faces under varying aspect views (pose) using a feature extraction procedure that inherently removes distortions due to pose variations, and therefore requires only single training and/or test face images, which could be at different aspect views.
Abstract: A new technique is discussed to recognize human faces under varying aspect views (pose). We first estimate the pose of an unknown human face from a 2D gray-scale image and then transform the unknown face to a reference pose using a feature extraction procedure. A different set of features for discriminating between different individuals are then extracted from these reconstructed faces for recognition. The feature extraction scheme used is known as the maximum representation and discrimination feature method. The advantage of our procedure is that it inherently removes distortions due to pose variations, and therefore requires only single training and/or test face images, which could be at different aspect views. For transformation, it does not require the face to be in the database during training. For recognition, only one aspect view at any pose is necessary.

Proceedings ArticleDOI
27 Sep 1999
TL;DR: Experimental results show that the proposed system for the analysis and automated identification of a human face is robust, valid for numerous kind of facial images in real scenes, works in real time with low hardware requirements and the whole process is conducted automatically.
Abstract: The paper deals with a system for the analysis and automated identification of a human face. A face can be recognized when the details of individual features are resolved. The idea is to extract the relative position and other parameters of distinctive features such as eyes, mouth, nose and chin. The overall geometrical configuration of face features can be described by a vector of numerical data representing position and size of main facial features. At first, from sequential images, eye coordinates are extracted by detecting eye winking. The interocular distance and eye position can be used to determine size and position of the areas of search for face features. In these areas binary thresholding is performed, the system modifies the threshold automatically to detect features. To find their coordinates, discontinuities are searched for in the binary image. Experimental results show that the proposed method is robust, valid for numerous kind of facial images in real scenes, works in real time with low hardware requirements and the whole process is conducted automatically.

Proceedings ArticleDOI
27 Sep 1999
TL;DR: A multi-modal database intended for person authentication from multiple cues, providing profile and frontal color images, 3D facial representations and many French and some English speech utterances is presented.
Abstract: This paper presents a multi-modal database intended for person authentication from multiple cues. It currently contains three sessions of the same 120 individuals, providing profile and frontal color images, 3D facial representations and many French and some English speech utterances. People were selected for their availability so that new sessions will be easily acquired. Individual recognition performances for speech, profile and 3D facial surface modalities are presented. The combination of these experts is the subject of current research.