InnovationQ will be updated on Sunday, April 29, from 10am - noon ET. You may experience brief service interruptions during that time.
Browse Prior Art Database

System and Method of a Hybrid Neural Network for Authentication with Facial Characteristics on a Camera-PDA Device

IP.com Disclosure Number: IPCOM000032318D
Original Publication Date: 2004-Oct-31
Included in the Prior Art Database: 2004-Oct-31
Document File: 3 page(s) / 70K

Publishing Venue



This article describes a new and innovative neural network system and method for authentication on PDA devices using the user's facical characteristics. The invention transforms and analyzes the unique facial signatures for authenticating the user identification. Once the user's facial characteristics is authenticated, the system unlocks the mobile device and allows the user to accessing his data and applications on the device.

This text was extracted from a PDF file.
At least one non-text object (such as an image or picture) has been suppressed.
This is the abbreviated version, containing approximately 53% of the total text.

Page 1 of 3

System and Method of a Hybrid Neural Network for Authentication with Facial Characteristics on a Camera -PDA Device

The invention is designed to use the image of the user caputured by the built-in camera on the mobile device to analyze the unique facial characteristics as authention signatures. The invention transforms the user's digital image and employs neural network technology to analyze the user digital image as credential inputs for authentication. There are 2 neural networks in the system.

The first neural network is to transform the digital image into a topological map which represent the unique characteristics of the user. The neural networks convert the digital image into authentication signatures in terms of image coefficient sets. This set of image coefficients is a unique representation of the user's digital image and is then used for authenticating.

The second neural network is used for classifying / matching of the user's credential. By using facial signatures to authenticating, this invention eliminate the tedious and inconvenient tasks of keying the user ID and password manually on the mobile devices.

System Architecture

At a high-level, the architecture of the system is shown in Fig. 1 below. It consists of the following components:

Digital Camera: A built-in VGA digital camera on the PDA.

Image Characteristic Analyzer : A software module for processing the digital image, transform and characterize the digital image into a 2-dimensional matrix of image coefficients Unique Feature Extractor : A software module for performing feature extraction of unique characteristics from the image coefficients. The unique image coeffients are used for (a) training the neural networks during authentication registering process; (b) inputs to the neural networks for authenticating.

Facial Characteristics Coefficient Neural Network : A neural network for training and forming credential coefficients.

Coefficients Authentication Neural Network : A neural network for authenticating user's credential.

User Credential Coefficients Database : A repository of user's neural network credential coefficients.

Credential Coefficient Validator : A credential matching algorithm.


Page 2 of 3

System Operations

The system is operated in 2 different processing modes: (1) Credential Registerin...