Latest question:
How do you think the new GigE standards will influence the machine vision industry?
Respond or ask your question now!
FEATURE
Long-Haul Camera Connections
For years, there were no off-the-shelf solutions for carrying camera data over long distances, and ad hoc solutions were clumsy at best. At last, an alternative may be just around the corner...
by Tom Morgan and Jack Okabayashi
February 2003

There are many scientific applications where it isn't practical to place a company's computer system at the same physical location as cameras or other sensor sources, astronomy being an extreme example. In industrial applications such as machine vision, the point of data capture frequently resides in an electrically noisy environment, so it's highly desirable to ship camera output to a remote computer for processing and storage.
There are, in fact, many situations where it's advantageous to move sensor data 100, 200 or 300 feet from the work area to a computer stashed on the other end of a production line. The problem is that there has been no off-the-shelf interface solution for handling the long haul. Standard camera interfaces, both legacy interfaces such as RS-422 and the new Camera Link interface, have severe distance limitations of well below 100 feet.
With that in mind, Bosque Computers is working with a new general-purpose digital interface from Inova Semiconductors, Inc., called GigaSTaR, in order to build systems capable of transmitting camera outputs over distances of 500 feet and beyond.
The Bosque digital video recorder is a 6-slot PCI system with 100 BaseT Ethernet, a frame grabber card, timing card, display card and RAID controller card, with a vacant slot for other off-the-shelf or custom cards customers might wish to add. The standard system also includes between 60 and 140 Gbytes or so of mass storage and a high-capacity tape backup. It's capable of capturing up to 500 frames/s of uncompressed data at 60 Mbytes/s, with custom versions capable of operating considerably faster. Although compressing sensor data can greatly miminize the transmission and storage burden on a system, uncompressed data provides the greatest fidelity and the best basis for subsequent image processing.