Signal Processing Toolbox 6.12

MathWorks Tools Crucial to the Development of Award-Winning Fingerprint Scanner

Challenge To build, test, and prototype a brand new line of computer security devices based on the concept of fingerprint recognition Solution Design and test the U.are.U™ fingerprint scanner using the analytical and visualization power of MATLAB, along with the specialized functions of the Image Processing and Signal Processing Toolboxes and the MATLAB Compiler Results An accelerated design process Fast, efficient prototyping An award-winning new product

Challenge

In 1996, engineers Serge Belongie and Vance Bjorn set out to produce a revolutionary line of computer security devices, built on the concept of fingerprint recognition. They knew that the highly competitive world of new product development would demand not only all of their considerable technical expertise but also a high degree of market savvy. If their new company, Digital Persona, was to succeed, Belongie and Bjorn had to sell their idea to investors. This meant producing a working prototype fast and then following through with technology that was both cutting edge and flawless. To do this, they needed to be able to test out their ideas quickly.

"MATLAB was indispensable for testing ideas, testing entire systems, and creating prototypes."

Serge Belongie
Digital Persona
Co-founder

Solution

Belongie and Bjorn developed their first product, called U. are U.™, using the analytical and visualization power of MATLAB along with the specialized functions of the Image Processing and Signal Processing Toolboxes.

U.are.U is a fingerprint identification system for personal computers and the Internet that is designed to interface easily with existing hardware and software. Unveiled in 1997, U.are.U includes a mouse-sized scanner with a USB connector and propriety software that scans and digitizes fingerprints, identifies authorized users and allows them quick computer access. The software takes four scans of a fingerprint then uses a set of proprietary algorithms, developed in MATLAB, to establish a print identification file that is used as a reference during each log-in attempt.

Belongie and Bjorn had discovered MATLAB during their studies at California Institute of Technology and Massachusetts Institute of Technology. According to the pair, MATLAB was so often an essential tool for their homework and labs that they became MATLAB "gurus." When Belongie and Bjorn joined forces at Digital Persona, they chose MATLAB for development and prototyping. "It was perfectly suited to the data structure of images and allowed us to test out ideas quickly," Belongie said.

"The addition of multidimensional arrays to MATLAB's feature set is particularly helpful for organizing image processing problems," he continued. "In one 3D array, I may store a set of 2D filters. In another, I may store a stack of images, starting with the raw image itself, proceeding through multiple filtered versions of that image, and concluding with an array of detected features. Never has it been so easy to literally 'look through the pipeline' of an image processing problem.

The Image Processing Toolbox also allows them to easily read and write various file formats, as well as resize and rotate images quickly and accurately. "I use the new PIXVAL command all the time," adds Belongie. "I in fact made it pop up by default any time I view an image." Also important to their work is the Signal Processing Toolbox, which they find "very useful" for exploring 1D and 2D filter designs.

Belongie and Bjorn use the MATLAB Compiler to generate C versions of their MATLAB functions (C-MEX files) to embed in their production code. The MATLAB Compiler allows them to test new ideas within the existing U.are.U™ code. "With the compiler, we can pop out the C function and insert a variation originally coded in the MATLAB language," explains Bjorn. "That way, the new algorithm, built in MATLAB, can quickly be tested with the rest of our production code and algorithms."

MATLAB will continue to play a central role in quality control on the optical and electronic components on Digital Persona's manufactured sensors. New projects in the works include making a standalone application with the MATLAB C/C++ Math Libraries that will inspect calibration images from the sensors and assess parameters such as blur, contrast, optical distortion, and lighting variations.

Results

Products Used