George Mason University

Harry Wechsler

Venkatesh Ramanathan

Shen-Shyang Ho

FaYin Li

Chengjun Liu

Hung Lai

Clifford Clausen

Hachim El Khiyari

Fayin Li

Time lapse, characteristic of aging, is a complex process that affects the reliability and security of biometric face recognition systems. Systems and methods use deep learning, in general, and convolutional neural networks (CNN), in particular, for automatic rather than hand-crafted feature extraction for robust face recognition across time lapse. A CNN architecture using the VGG-Face deep (neural network) learning produces highly discriminative and interoperable features that are robust to aging variations even across a mix of biometric datasets. The features extracted show high inter-class and low intra-class variability leading to low generalization errors on aging datasets using ensembles of subspace discriminant classifiers.

Age invariant face recognition using convolutional neural networks and set distances

A new robust human authentication system, device, and instructions, embeddable in a physical and tangible computer readable medium, for determining if at least one test image obtained using an imaging device matches at least one training image in an enrollment database, are disclosed. This invention applies the concepts of appearance (PCA or PCA+LDA) and holistic anthropometrics that include head, face, neck, and shoulder linear and non-linear geometric measurements. The appearance ('eigen') coefficients and holistic anthropometric measurements selected may be used as feature vectors. A boosting algorithm ranks features as 'weak learners' and combines their outputs for &#x201c;strong&#x201d; recognition.

Robust human authentication using holistic anthropometric and appearance-based features and boosting

A robust recognition-by-parts authentication system for comparing and authenticating a test image with at least one training image is disclosed. This invention applies the concepts of recognition-by-parts, boosting, and transduction.

Face authentication using recognition-by-parts, boosting, and transduction

A data stream change detector including a receiving module, a preprocessor module, a clustering module, a strangeness module, a p-value module, a martingale value determination module, comparison module, and an output module. The receiving module accepts new data vectors that originate from a sequence of data in a data stream. Preprocessor module preprocesses the new data vector using a filter. The clustering module clusters the new data vector with previously received data vectors. Strangeness module computes a strangeness value for each of the previously received data vectors. The p-value module calculates a p-value for the new data vector using the strangeness value. Martingale value determination module calculates a martingale value for the new data vector using the p-value. Comparison module compares the martingale value with a threshold value; and sets an indicator if a change if the martingale is greater than the threshold.

Data stream change detector

A recognition-by-parts authentication system for determining if a physical test target represented in test image(s) obtained using an imaging device matches a physical training target represented in training image(s). The system includes a multitude of adaptive and robust correlation filters. Each of the adaptive and robust correlation filters is configured to generate correlation-peak-strength and distance-from-origin data using a multitude of related images. Each of the multitude of related images representing a similar part of a larger image. The related images originate from the test image(s) and training image(s).

Recognition by parts using adaptive and robust correlation filters

An open set recognition system utilizing transductive inference including capture device(s), a basis, quality checker(s), feature extractor(s), a gallery, a rejection threshold, a storage mechanism, and a recognition stage. The basis encodes sample(s) and is derived using training samples. The feature extractor(s) generates signature(s) from sample(s) using the basis. The rejection threshold is created using a rejection threshold learning mechanism that calculates the rejection threshold using sample(s) by: swapping a sample identifier with other sample identifier(s); computing a credibility value for the swapped sample identifiers; deriving a peak-to-side ratio distribution using the credibility values; and determining the rejection threshold using the peak-to-side ratio distribution. The open set recognition stage authenticates or reject as unknown the identity of unknown sample(s) using derived credibility values, derived peak-to-side ratios for the unknown sample and the rejection threshold.

Open set recognition using transduction

A method and system for determining the similarity between an image and at least one training sample is disclosed. This invention is a novel Gabor Feature Classifier (GFC), a principal application of which may be for face recognition. The GFC applies the Enhanced FLD Model (EFM) to an augmented Gabor feature vector derived from the Gabor wavelet transformation of images.

Feature based classification

A method for fractal image compression using reinforced learning is disclosed. The reinforced learning algorithm improves the performance of fractal image compression algorithms by improving image partitioning and transform family selection while considering the impact of domain to range matching. The present invention differs from other fractal image compression algorithms in that it makes decisions about transform family and image partitioning by generalizing from experience compressing small portions of an image to compressing an image as a whole.

Growing community of inventors

Fairfax Station, VA, United States of America

Harry Wechsler