Catálogo de publicaciones - libros

We present results on classification of palmprint patterns from a large number of classes for biometric verification. We train optimal trade-off correlation filter classifiers with patterns of subregions of the palm as the actual biometric for the person’s identity. Our results show that with less than 5 cm (less than 1 in) of the actual palm captured at a low resolution, correlation filter algorithms can verify the authenticity of the palmprint pattern with error rates below 0.5% from as many as 400 different patterns. There is no previous work on biometric palmprint recognition that studies pattern verification of such small palmprint regions with such large number of classes.

This paper presents the evaluation of face recognition performance using visual and thermal infrared (IR) face images with advanced correlation filter methods. Correlation filters are an attractive tool for face recognition due to features such as shift invariance, distortion tolerance, and graceful degradation. In this paper, we show that correlation filters perform very well when the face images are of significantly low resolution. Performing robust face recognition using low resolution images has many applications including human identification at a distance (HID). Minimum average correlation energy (MACE) filters and optimal trade-off synthetic discriminant function (OTSDF) filters are used in our experiments showing better performance over commercial face recognition algorithms such as FaceIt® based on Local Feature Analysis (LFA) using low resolution images. We also address the problems faced when using thermal images that contain eyeglasses which block the information around the eyes. Therefore we describe in detail a fully automated way of eyeglass detection and removal in thermal images resulting in a significant increase in thermal face recognition performance.

The iris is considered one of the most reliable and stable biometrics as it is believed to not change significantly during a person’s lifetime. Standard techniques for iris recognition, popularized by Daugman, apply Gabor wavelet analysis for feature extraction. In this paper, we consider an alternative method for iris recognition, the use of advanced distortion-tolerant correlation filters for robust pattern matching. These filters offer two primary advantages: shift invariance, and the ability to tolerate within-class image variations. The iris images we use in our experiments are from the CASIA database and also from an iris database we collected at CMU. In this paper, we perform automatic segmentation of the iris (which surrounds the pupil) from the rest of the eye, normalizing for scale and pupil dilation. We then use these segmented iris images to compare the recognition performance of various methods, including Gabor wavelet feature extraction, to correlation filters.

In this paper, we propose a secure and efficient protocol to transmit fingerprint images from a fingerprint sensor to a client by exploiting characteristics of fingerprint images. To guarantee the integrity/confidentiality of the fingerprint images transmitted, a standard encryption algorithm is employed. Because the fingerprint sensor is computationally limited, however, such encryption algorithm may not be applied to the full fingerprint images in real-time. To reduce the computational workload on the resource-constrained sensor, we apply the encryption algorithm to a nonce for integrity and to a specific bitplane of each pixel of the fingerprint image for confidentiality. Experimental results show that the integrity/confidentiality of the fingerprint images can be guaranteed without any leakage of the ridge information. Also, the image-based selective bitplane encryption can be completed in real-time on embedded processors.

Biometric authentication has been considered a model for quantitatively establishing the discriminative power of biometric data. The dichotomy model classifies two biometric samples as coming either from the same person or from two different people. This paper reviews features, distance measures, and classifiers used in iris authentication. For feature extraction we compare simple binary and multi-level 2D wavelet features. For distance measures we examine scalar distances such as Hamming and Euclidean, feature vector and histogram distances. Finally, for the classifiers we compare Bayes decision rule, nearest neighbor, artificial neural network, and support vector machines. Of the eleven different combinations tested, the best one uses multi-level 2D wavelet features, the histogram distance, and a support vector machine classifier.

This paper proposes an algorithm detecting facial component to efficiently extract the FCP (Facial Characteristic Point). The FCP plays an important role in facial expression representation or recognition. For efficient FCP extraction using image processing, we analyze and improve the conventional algorithms detecting facial components that are the basis of the FCP extraction. The proposed algorithm includes face region detection without the effect of skin-color hair, eye region detection with weighted template, eyebrow region detection using a modified histogram, and mouth region detection using skin characteristics.

A new approach for recognizing facial expressions in various internal states that is illumination-invariant and without detectable cues such as a neutral expression is proposed. First, we propose a zero-phase whitening step of the images for illumination-invariant. Second, we developed a representation of face images based on principal component analysis(PCA) representation excluded the first 1 principle component as the features for facial expression recognition, regardless of neutral expression. The PCA basis vectors for this data set had reflected well the changes in facial expression. Finally, a neural network model for classification of facial expressions based on dimension model was created. The dimensional model recognizes not only six facial expressions related to six basic emotions (happiness, sadness, surprise, angry, fear, disgust), but also expressions of various internal states. PCA representations excluded the first 1 principle component and neural network model on the two-dimensional structure of emotion have improved the limitation of expression recognition based on a small number of discrete categories of emotional expressions, and have overcome the problems of lighting sensitivity and dependence on cues such as a neutral expression.

Facial feature extraction is a fundamental problem in image processing. Correct extraction of features is essential for the success of many applications. Typical feature extraction algorithms fail for low resolution images which do not contain sufficient facial details. In this paper, a region-based super-resolution aided facial feature extraction method for low resolution video sequences is described. The region based approach makes use of segmented faces as the region of interest whereby a significant reduction in computational complexity of the super-resolution algorithm is achieved. The results indicate that the region-based super-resolution aided facial feature extraction algorithm provides significant performance improvement in terms of correctly detecting the location of the facial feature points. There are 6.4 fold reductions in the computational cost.

In this paper an intelligent and efficient combination of several methods are employed for face and facial feature tracking with the motivation for real time applications. Face tracking algorithm is based on color and connected component analysis. It is scale, pose and orientation invariant, and can be implemented in real time in controlled environments. The more challenging problem of facial feature tracking uses intensity based adaptive clustering on facial feature sub-images. New search region estimation for each sub-image is proposed. The technique employs facial expression aware eye sub-image prediction. The simulation results indicate that facial feature tracking is efficient with an average tracking rate of 99% with a three pixel range under different head movements such as translation, rotation, tilt, and scale changes. Furthermore it is robust under varying facial expressions and non-uniform illumination.

There has been many hidden communication techniques proposed in the last few years. The focus was given to steganography to build such techniques. Utilizing stego-key(s) to hide secret messages into images strengthen the security of these techniques. However, adopting one of the available key-agreement protocols, to distribute stego-key(s) between the communicating parties, will destroy the infrastructure of the entire communication. The reason is that, these protocols perform their transactions on sight, while the desirable communications need to be completely hidden. In this paper, a is proposed to be added to the steganography general model. This unit utilizes a new key-agreement protocol, , to help support the entire class of hidden communication techniques to exchange the sego-key(s) covertly. The proposed stego-KA protocol is based on key establishment protocol and has significant advantages that support hidden communications.