Faster processing empowers imaging applications
George Kotelly
Executive Editor [email protected]
Dramatic improvements in microprocessor technology and imaging algorithms are enabling the development of advanced imaging applications. By meeting the high-speed demands of custom software, faster and more powerful microprocessors have been able to accelerate image processing and analysis. The integrated hardware and software has resulted in expanded imaging applications such as for personal identification, pattern recognition, and medical-data integration.
In personal identification systems, a fast-growing science called "biometrics" has been transformed from a statistical science to practical applications. In the transition process, it has spawned a variety of identification and verification techniques, such as retinal scans, digital signatures, finger imaging, and face recognition. These techniques have emerged, says contributing editor Shari Worthington, because of the marked improvements made in new microprocessor technology (see p. 20).
Pattern-recognition applications have forced available image-processing systems into complicated trade-offs in hardware, development costs, maintenance of training sets, and accuracy. To overcome these limitations, companies are turning to approaches such as including neural networks and fuzzy logic. By incorporating both of these approaches into its video-recognition system, states technical director Charles Hooge, BrainTech Inc. has developed a PC-based system for extracting meaning and recognizing objects in a stream of video data (see p. 32).
Called TeleMed, a nationwide patient-record database system has been designed that allows clinicians, diagnosticians, hospital residents, and physicians to access medical information from their homes or offices via a multimedia interface. According to staff editors George Kotelly and Andy Wilson, through the use of object-oriented programs, medical experts can view repository data simultaneously, use it to collaborate and consult over a wide-area network, and then determine course of action (see p. 26).
To deliver the necessary information for processing machine-vision and imaging systems, designers are confronted by hundreds of off-the-shelf data-acquisition boards from more than 50 vendors. Designers must therefore wrestle with diverse board types, notes staff editor Andy Wilson, and must also carefully evaluate differing operating systems, throughput, scalability and vendor support, to meet system needs (see p. 38).
