How do you think the new GigE standards will influence the machine vision industry?
Respond or ask your question now!
LEARNING BY DOING
The concept of embedding a self-learning machine vision system into a high-performance digital video camera may seem like a fish story, but not to the folks at PULNiX America. When members of the commercial fishing industry needed a more efficient way to sort mackerels from their daily catches, PULNiX provided a solution with vision.
The ZiCAM Zi-640 smart camera uses machine vision to perform repetitive, learnable tasks-such as sorting fish-that would otherwise require human supervision. Integrated into this camera is a neural network-based, 312-neuron Zero Instruction Set Computer (ZISC), designed to execute image-understanding algorithms without the need for programming.
Like a child learning to fit different-shaped blocks into corresponding holes, the Zi-640 learns in time what a user needs it to do in a given application. The camera can then operate independently as a stand-alone system-and in real-time, at a rate of 30 frames/s.
Available in both monochrome and color, PULNiX's ZiCAM has a direct monitor drive providing real-time video without support electronics. An integral 340MB microdrive makes it possible to store raw video images-even those captured in so-called "no-go" conditions. This patented 640 x 480 pixel digital camera can handle applications where humans performing Quality Control via Statistical Processing Control (SPC) might normally grow fatigued, leading to missed defects. DVT, Cognex and other competitors offer similar products.
Since the Zi-640 can mimic human learning, PULNiX recommends it for such applications as presence/absence, inspection, positioning and recognition. This, according to Don W. Lake, PULNiX's Director of Marketing, makes it especially suited for the pharmaceutical, manufacturing, and security industries-not to mention commercial fishing-where locating a specific Region of Interest (ROI) is vital to success.