System integrators hurdle performance barriers
George Kotelly, Executive Editor
The integration of hardware, software, and optics into a task-oriented, operational machine-vision or imaging system can be complex, time-consuming, and costly. This month`s Vision Systems Design describes several vision systems in which integrators were profoundly challenged to improve performance.
For example, machine-vision designers overcame the burden of improving the viewing and speed functions of robotic systems by incorporating additional cameras and sensors. According to science writer Dave Wilson, three cameras instead of just one are being used to inspect surface-mounted devices in the x, y, and¥planes, and vacuum sensors are verifying parts selection and placement (see p. 16). In this approach, system multitasking speed is boosted as the sensors check for the presence of the next part as the robot is placing the previous part.
For added food-inspection discrimination, researchers at the University of Saskatchewan in Canada have developed a system to evaluate the quality of popcorn kernels by using off-the-shelf hardware, lighting systems, and software (see p. 28). By integrating a personal computer, memory, frame grabber, color camera, and neural-network software, their system enhances image acquisition and identifies and quantifies kernel features too small for the human eye to see.
Neural-network algorithms are also improving the accuracy and reliability of fingerprint-identification systems, says contributing editor Richard Parker (see p. 34). These algorithms compare new camera images to computer database images and deliver this information to a decisioning stage, which has been trained by a probabilistic technique.
Aerospace applications are also benefiting from refined system integration. The NASA Mars Global Surveyor spacecraft has completed its first month of studying the Martian surface composition, atmosphere, and magnetic properties, reports contributing editor John Haystead (see p. 22). On board is a spectrometer that conducts a survey of mineral composition, cloud formations, atmospheric dust, and weather patterns.
In our Product Focus, contributing editor Rick Nelson finds that systems integrators are confronted by daunting specification choices in selecting color displays. Resolution, size, viewing area, dot pitch, shadow-mask and aperture-pitch tube technologies, and decreasing prices are criteria that determine the proper color monitor for the application (see p. 40).
