Different software for different tasks
Jon Titus, Contributing Technical Editor -- Test & Measurement World, 10/1/2003
The area of "vision software" can get confusing. On one hand, engineers use vision software in inspection applications that make real-time decisions to accept or reject a product. Typically, inspection systems locate fiducial marks (targets), compare images, identify shapes, and report dimensional quantities. Think of a system that visually checks for the proper placement of SMT components on a PCB prior to reflow.
On the other hand, vision software performs image-analysis functions that emphasize changes, remove noise, sharpen edges, and enhance features in an image. Think of manipulating x-ray images during failure analysis to enhance defects on a BGA's solder balls. In general, that sort of operation doesn't require real-time processing.
![]() |
|
Machine-vision tools solve specific real-time problems such as finding edges and detecting patterns. Courtesy of National Instruments. |
A common set of tools may save time and money, or so it seems. But this apparent economy usually proves false. Software suppliers take pains to optimize their tools for specific applications, so even though image-analysis tools may handle inspection tasks, that doesn't mean they should, or that they can handle them well.
For the most part, developers should use the tools meant specifically for a given type of application. Try to apply an image-analysis tool to an inspection task and you may lose real-time capabilities, and control of I/O devices may require extensive software development. Likewise, trying to use a machine-vision tool to perform a complex image transformation may lead to coding nightmares.
There's another aspect that developers also must consider: support. It's unlikely the vendor of one type of software will offer extensive application assistance, supporting software, add-on hardware, and training in unfamiliar areas. Ask machine-vision experts about an image-analysis operation such as a dipole sheet transform, and they'll scratch their heads. So, although the idea of adapting one set of vision tools to disparate tasks looks intriguing, you'll find that these tools, like all tools, apply best to their intended jobs.
Jon Titus, Contributing Technical Editor, jontitus@attbi.com
|
















The compact and lightweight Omega infrared camera from Indigo Systems now comes with a Firewire (IEEE 1394) interface. The camera also can produce an RS-170 video signal or a 14-bit digital output. The camera requires no internal thermoelectric cooler, and as a result, it consumes only 1.5 W. 


