Smart cameras serve as LabView targets
Rick Nelson, Chief Editor -- Test & Measurement World, 1/29/2008 9:30:00 AM
National Instruments’ monochrome VGA NI 1722 and NI 1742 smart cameras have 400-MHz and 533-MHz PowerPC processors, respectively. NI Vision product manager Matt Slaughter commented on NI’s entry into the smart-camera market in an exclusive interview.
Q. When did NI introduce the two smart cameras?
A. We had a preannouncement at NIWeek last summer, where we showed off some of the features during a keynote demo, but the official announcement occurred at Vision 2007 in November.
Q. NI is reselling some third-party cameras—why not take that approach with the smart cameras?
A. We have GigE Vision and 1394 [FireWire] cameras from Basler that we are reselling, and there certainly are other smart camera vendors out there—some of whom ship their cameras with our software. But we wanted to make sure that our smart cameras represent a true LabView target that we had complete control over.
GigE Vision and 1394 cameras comply with set standards, so regardless of what third-party camera we are using, our back-end software can stay the same. That’s not the case with smart cameras, and we had to do a lot of driver development to get our smart cameras to work with all the software we have.
Q. Can you give an example of these cameras’ compatibility with NI software?
A. For NIWeek, we like to do run-throughs a month early to make sure everything goes smoothly, and I didn’t have a new smart camera available to use to develop the demonstration. So, I used a monochrome analog camera with the same 640x480-pixel resolution. I wrote my entire application in LabView using a standard driver, and I was able to directly port that application over to the smart camera with minimal changes.
Q. Your smart cameras have two Ethernet ports. Are they GigE Vision compliant?
A. No, we are emphasizing that they are not GigE Vision cameras. One port will be used for reporting results and maybe for reporting failed images back to your main host. The other port is typically going to be used for one of two things: The first is troubleshooting, in which you can walk up and plug in a laptop via crossover cable and do troubleshooting without ever pulling the camera off the line. The second thing that most people will probably do with this second port is communicate with other devices, like PLCs or our CompactRIO platform.
Q. Will you add GigE Vision in the future?
A. That’s something that’s been brought up a few times. But the purpose of a GigE Vision camera is of course to return images as quickly as possible, and you typically would not go with a $2000 smart camera to do that. So, I don’t think we get much value added by turning this into a GigE Vision camera. But I’m not going to say it will never happen.
Q. The NI 1722 and NI 1742 represent NI’s first foray into the smart camera market—is it the last?
A. No, it’s not just these two cameras we are releasing; it’s going to be an entire family of cameras. You can expect to see an announcement about every quarter.
Q. How do the NI 1722 and NI 1742 smart cameras complement NI’s other vision products?
A. They fit very well within the product line that we currently have, which ranges from our Compact Vision System up to PC-based systems. With all these products, including the new cameras, you have two development options: configuration-based, using our Vision Builder software, or programming-based, using LabView. We are also able to use the shared-variable technology that’s been around since LabView 8.0 and that we’re now fully supporting within our vision platform, so we can easily communicate shared variables between our vision platform and the rest of NI’s control platform.
Q. Why are you using CCDs instead of CMOS?
A. In a lot of these lower-end, low-cost industrial applications, you’re seeing a lot of CMOS sensors because they’re a little cheaper to make. We went with the CCD, because you tend to get a little bit better image quality.
If you take a look at the images we get out of our cameras vs. what you get out of some of the other cameras on the marketplace that are in our same price range, you’ll notice a drastic difference in the two. And of course, with a sharper image you are able to do things like edge detection and pattern matching much more effectively.
We wanted to make sure you get a really good crisp image to work with, and the CCD seemed to be the best way to go for that. The cost penalty was not significant.
Q. What could the cameras’ optoisolated digital I/O be used for?
A. Well, you can actually do pretty much whatever you want with it. As one example, our cameras can generate modulated pulse trains, which will let you do some interesting applications.
At NIWeek, we had a demo where we were inspecting beer bottles. In this demo, we were not only checking to see if the label on each bottle was correct, but we also aligning each bottle with a small stepper motor. And we could use the camera’s pulse output to control the stepper motor without requiring a PLC or Compact RIO module.
Q. NI offers two speed versions. How do I choose between them?
A. The processing time for an image its really dependent on what processing you are doing—pattern matching and image rotation are of course slower operations. But functions like edge detection are very quick operations.
So, for the applications out there where you need to do these higher-processing-power applications like pattern match or optical character recognition and you need to do it quickly, the 400-MHz version of the camera may not do everything you need, so that 533-MHz version will provide you with more power.
Q. Should I develop my software first and then choose the camera?
A. That’s a very good way to go about it. We provide a performance meter that you can run so you can see exactly how much processing power you have left on your system as you do your development. Also, your application might evolve over time.
For example, your system might run for six months or a year, and then someone changes the requirements by adding a new tag on a bottle or upgrading a conveyer to run 20% faster. If you’re using the 1722 and suddenly need to get a little more juice out of the camera, then you can simply move up to the 1742.
Q. The 1742 has a built-in lighting controller—is that a common feature?
A. That’s actually a very uncommon feature in smart cameras. Engineers who use this feature won’t have to go buy that external lighting controller that’s going to cost $500 to $1000, and they don’t have to worry about all the extra wiring and the extra programming. With our smart camera, the API [application programming interface] for the lighting controller is built directly into the image-acquisition API for the smart cameras.
Q. The 1742 also has a quadrature encoder feature. What’s that used for?
A. The quadrature encoder input is useful for those types of applications in which you are controlling a stepper motor and you want to make sure that you get the stepper in the exact correct location every time. All you have to do is provide the quadrature encoder feedback off the stepper motor into the 1742. The quadrature encoder is a really nice feature to have if you want to do some really precise timing on image acquisition on a belt-driven system that’s connected to a liner drive or a rotary drive.
www.ni.com/vision
Talkback
Related Content
Sponsored Links
TMW Resource Center
Browse Resources by Type:
- PCI Express Advances Machine Vision
DALSA | White Paper
VIEW NOW
DALSA | White Paper
VIEW NOW
DALSA | White Paper
VIEW NOW
View More
