New software naming convention?

I don’t think anyone would argue that Cognex’s PatMax tool is both powerful and capable. When you need to locate a part in an image and your regular fixturing isn’t reliable or becomes too complex PatMax will probably do it.

But, (there’s always a “but” in this blog,) it is computationally intense. Matching contours takes a lot of digital horsepower, which results in extended processing time, especially when running PatMax on a high-res InSight smart camera. So I sat up and paid attention when I spotted a recent Cognex press release.

“New Cognex Feature Location Technology Offers Unmatched Performance and Speed” (May 5^th, 2015) announces “PatMax RedLine™, a feature-location technology that reinvents the industry-leading PatMax® pattern-matching tool by maximizing speed and performance.” What this does, or so Cognex claim, is speed up the way PatMax runs on a 5Mp InSight.

I’m sure that’s a good thing, and I suspect, although it’s not mentioned in the release, that Cognex charge a premium for its use. However, it does beg two questions:

1. Why didn’t my local Cognex sales rep let me know this was coming?
2. ‘Redline’ sounds like a subway system, (Boston? San Francisco?) in which case I wonder if we should brace ourselves for ‘District’, “Piccadilly’, and ‘Bakerloo’?

Well if Google can use weird naming conventions for Android releases, what’s stopping Cognex from doing the same?

Smart camera or PC vision?

It’s one of those eternal questions, so I’ve put together a little graphic. I hope this will at least prompt you to think about what matters most in your application and guide you making the correct decision. (There’s no right or wrong, it’s just a case of what works best for you.)

The five axes I chose are:

• ease of support – how easy can a system be maintained?
• implementation speed – can you install it quickly?
• task complexity – do the images need intensive processing?
• data rate – a combination of image size and frame rate
• number of cameras – dumb cameras are inexpensive, smart ones are dear

Based on this admittedly rough-and-ready analysis, I think you can see that:

• PC-based vision wins when the task is complex or you need multiple cameras
• Smart cameras should be your first choice when you need rapid installation and ease of support and the task is not too complex.

Smart cameras are old-school?

I’ve made the argument that PC-based machine vision can be less expensive smart cameras. That’s especially true when resolutions go up and more than one camera is needed. I know some of you agree, but I’ve never heard it enunciated quite so clearly as Datalogic’s Gary Koken does in “Changing Asian Economics Sew Seeds for Automation Growth in Textile Industry” (Vision Online, July 1^st, 2013.)

The top half of the article you can skip since it’s all about the Chinese textile industry, but part way down you’ll come to “PCs Dominate Asian Machine Vision” and that’s where you should start reading.

This is where Koken explains the cost issue. I’ve no argument with his numbers, (well perhaps his 2Mp camera prices are a little steep,) but what I don’t see is any data to support that subheading.

I think the smart camera versus PC solution is a whole lot more complex than just the purchase price. There’s speed of implementation, training, support, usability, standardization, upgrade pathways and more to consider. In fact there’s so much to take into account I see a whole series of blog posts on the subject…

Smart camera upgrades from Matrox

Say “smart camera” and most machine vision people will respond “Cognex”. My guess is that only one in fifty would go “Matrox” and that’s a pity because Matrox has a very viable smart camera product.

Going by the name of Iris, and more recently Iris GT, it’s a capable product that runs a software product they call “Design Assistant” or DA for short. Now I’ve played with DA and I found it good, but not so good that I was filled with a burning desire to use it for every application.

But I hear there’s a pretty significant upgrade coming to Iris and DA. Things like more memory in the cameras, (always a good move,) upgrading the camera OS, and a host of new and improved tools for DA. You’ll need to wait for the official announcement from Matrox to learn what these are, but it’s my understanding that a filmstrip image replay function, like that on the Cognex Checker, is on the way.

Improvements are good, even if it is a bit of a pain-in-the-ass to switch over, but I’m still not sure this will convert me to the DA cause. But maybe I should download the evaluation version and give it a go.

Smart camera processors

Interesting article on the VisionOnline website for those curious about the horsepower in their smart cameras. “Smart Cameras Deliver PC-Host Power” (July 25^th, 2012,) discusses trends from the earliest days of smart cameras to the present. The message is that smart cameras are rapidly acquiring full PC capabilities, which is why they can run products like Halcon, but it wasn’t always so.

One of the biggest issues has been heat, which is where the Intel Atom began making inroads. And now, with the latest generation of i5 and i7 processors finding their way in to “ultrabooks” I have to think they’ll soon be hooked up to cameras too.

What, I wonder, are the implications of this shift? Will machine vision users want Windows-based smart cameras, or will we migrate to Linux? I can see that becoming an attractive path, in which case Ned Lecky of Voyant Vision will be on to another winner. Or is there a case for Google’s Chrome?

Just a thought.

Am I the last to know?

Another one to file under “why wasn’t I told?” – Festo, a name I associate with pneumatics, also offers vision sensors. If you take a look at their website, under Products – Image Processing you’ll find the SBOI and SBOC smart cameras. A glance at the specs suggests these are reasonable enough products, the differences being that the SBOC runs up to 1.3Mp resolution and takes a C-mount lens. But on the same web page there’s also a product called “Checkbox”.

Being the curious type, I took a look, and blow me if it isn’t a linescan smart camera! The Festo website isn’t particularly friendly, so it’s hard to learn much about this product – actually two products because there’s Checkbox and Checkbox Compact. Also lacking is much in the way of application case studies, so all told, it’s hard to figure out just how capable a product this is.

Now I imagine Festo isn’t actually manufacturing these products – the SBOI in particular looks quite familiar – so where does the Checkbox come from? And why haven’t I heard about it before?

New vision sensors and systems

I must be awfully busy these days, because I missed the announcement from Baumer regarding their Verisens® vision sensors. Now you may be asking if the machine vision world really needs another sensor, so let me explain why I think this is an interesting new family of products.

Technically, there’s no new ground broken, but I like how one package offers everything from bar code reading all the way up scale of vision complexity to full part inspection. Now that’s not all in a single box, mind you. At the bottom of the range is the ID-100 for code reading. Then there’s the CS for presence/absence tasks, followed by the XF and XC for inspection.

These last two models incorporate Baumer’s FEXLoc® pattern-matching technology, so they have the ability to search for objects in an image. The XC goes a step further still by using a C-mount lens connection and offering sensors up to 2Mp. (The lesser models use 10 and 16mm lenses.) In other words, it’s a fully-fledged smart camera. The only omission from the range is any color capability, but I imagine that will come soon.

So in effect, the Verisens sensor appears to cover the Dataman, Checker, and InSight families from Cognex. And one of my pet gripes about the Cognex products is that they’re all stand-alone. In other words, you can’t, (or I haven’t figured out how to,) move an application from one product up to the next. With Verisens it’s my impression that upgrading would be considerably more straightforward.

So there you have it: an interesting vision sensor to smart camera product line from Baumer. Take a look.

Under the hood of the Ximea Currera-G

Since bursting on the machine vision scene a couple of years ago camera-maker Ximea seems to have gone rather quiet. I imagine that’s because they’ve been busy selling cameras, although as this blog post by AMD suggests, they’ve also been working on new products.

Their Currera camera range is essentially a PC smart camera, which means it can run whatever software you prefer. However, as the AMD post tells us, the first generation used an Intel Atom processor. There’s nothing wrong with that, but I speak from personal experience when I say that while its energy consumption may be low, it ain’t that fast.

So perhaps it’s not surprising that the second-gen Currera G uses an AMD processor. According to AMD this results in a 45 times speed increase. I’m thinking that makes the constraint in the trigger-to-result pipeline the speed at which pixel values can be shuttled off the imaging chip rather than the processing of those values. And given that that will be much faster than 90% of applications demand, there’s potential either to do lots of intensive grayscale correlation pattern matching, or to acquire bigger images.

It will be interesting to see what Ximea do with all the extra horsepower.

What can you do with Halcon?

The simple answer is – pretty much any machine vision task can be done with Halcon from MVTec. It’s one of the world’s leading vision software packages, although I don’t see it a lot in North America. (If you have a good Halcon application, please let me know.)

So why isn’t everyone using it? I think there are three barriers.

First, it’s not simple to learn and use. I’m sure MVTec will email me to explain why I am wrong, but in my defense let me say that I have looked at it, I am impressed with it, but with my bias towards a graphical programming interface I see one heck of a learning curve. More of a learning cliff really, think north face of the Eiger. Of course, to cater to folks like me MVTec also offer ActivVisionTools.

The second barrier is price. This is not inexpensive software.

The third barrier is, I believe, a simple lack of knowledge of what one can do with Halcon. This is the one I’m going to put right.

I recently stumbled across an excellent machine vision blog written by Stéphane François of Ontario-based Computed Vision Solutions. Stéphane is a Halconist. I’m not sure if that’s a formal designation like being a Certified LabVIEW Developed or just a title one bestows on oneself, but it’s clear that he knows his Halcon inside out and from front to back.

Stéphane seems to have a little time on his hands – or just manages his time well – because on his blog, titled simply, “Computed Vision Blog”, he describes, in great detail, some of his explorations in machine vision. This year he’s investigated how to run Halcon on a Vision Components nano smart camera, and is now getting to grips with the 3D SICK Ranger product.

There are two things I really like about Stéphane’s blog. First, there’s no BS. He just writes about what he does and observes, with lots of pictures to help explain. And second, he posts lots of video. So if you’re at all interested in learning what Halcon could do for you, (or you want to learn about the VC nano or the SICK Ranger,) log on and get reading.

High dynamic range smart camera

While searching through the website of IMAGO Technologies I noticed that they offer a smart camera that uses the same LinLog dynamic range-extending technology as found on some of the PhotonFocus cameras. (Products – then scroll down to Intelligent Industrial Cameras, and click on VisionCam PS.) I’m excited about this, (I excite easily,) because I’m very impressed by LinLog, and like the idea of having that capability in a smart camera.

If you’re interested in learning more about what LinLog can do I suggest you visit the PhotonFocus website, but for now let me say that it lets you image very bright things and things that are not bright at the same time. Like the filament of a light bulb when it glows, or a weld pool.

Interesting things going on at IMAGO.

Vision-based tool protection

I don’t want you getting the idea that I’m secretly working for Teledyne Dalsa, but here’s another interesting story about an application of their BOA smart cameras. It comes from the February 2012 pages of Control design magazine and is titled, “Avoid a Pressing Problem”.

Most smart camera applications seem to be for product quality control, but in this case the goal was to protect expensive tooling and prevent unplanned downtime. You can read the article for the details; my goal here is to open your eyes to a whole new class of vision-based cost reduction opportunities.

With mold tooling costing tens of thousands of dollars and equipment utilization paramount in hitting cost targets, I’m surprised more effort isn’t put in to these types of applications. Of course, it’s often very difficult to get a good view of the tooling, and the environment can be pretty camera-unfriendly, but I’ve long wondered about some kind of scanner – one of those CIS-type sensors perhaps – that would check everything is shipshape with the tooling between cycles.

This BOA system doesn’t get that fancy, but it shows it can be done.

New Cognex smart cameras

My Cognex sales rep doesn’t love me any more. If he did, he’d surely have told me about the new 7000 series of In-Sight smart cameras, maybe even brought me one to play with it. As it is, I would never have learnt about this new range if I hadn’t been wasting my working hours on the web.

What’s neat about these cameras is the frame rates. While 30fps has long been the norm for VGA, and a 2Mp camera might get up to 15frames, the 7000 range trips along at over 100 frames per second from a 600 x 800 sensor. (I have to assume the sensor is a CMOS chip, although that doesn’t concern me: these days I find CMOS quality perfectly adequate.)

So why do I like high frame rates? Well some production lines just run that fast, some faster, but I also like how high frames can give you several chances to see. Particularly if a part is in motion, high frame rates make it possible to grab multiple images under lighting conditions that change subtly. This can be a great way of checking features that lack really good contrast. Another possible application is for linescan replacement. Snap away at a web fast enough and you can perform 100% surface inspection.

No word on pricing – perhaps that’s why my sales rep is keeping his distance: he knows I’ll want to beat him up a bit because the 7000 series In-Sights are probably expensive.

My theory on smart cameras from NI

This follows on from my post “New smart cameras from NI” so if you haven’t read that, click the link and do so.

Now, answer me this: why would NI introduce a new range of smart cameras with specifications remarkably similar to those of the Iris GT from Matrox? And if you don’t think they’re close perhaps you’d like to highlight one area of difference, because I can’t find it.

Here’s what I think is going on. NI want to be in the smart camera business as a way of pushing their very capable VBAI vision software. Unfortunately neither VBAI nor their range of low cost smart cameras have set the world alight, so how do they maintain a presence while minimizing costs?

They buy cameras from someone else.

This then is an interesting trend: the smart camera hardware becomes something of a commodity item, the differentiation comes from the software the user selects.

I have no problem with this, in fact I suspect it’s a trend we will see more of: vision companies will focus on the software that differentiates their products while one or a few companies will remain in the hardware manufacturing business.

Is this machine vision consolidation?

Interesting news out of Germany: camera manufacturer NET (short for New Electronic Technology) is buying image processing experts iv-tec. I’ve been predicting consolidation among camera makers for some time, but this strikes me as being more like vertical integration.

NET’s niche has really been FireWire cameras, although they do sell the Eyespector smart camera family. Meanwhile iv-tec have some serious expertise in digital image processing, as evidenced by these movies on their web site. Machine vision cameras, and especially smart cameras, of course rely on image processing software, so there is some logic to the acquisition. In fact it wouldn’t surprise me in the least to learn that these two companies have been working together for some time, but where is this leading?

Well here’s my hypothesis: iv-tec are doing some interesting things in the field of 3D, especially with stereo image processing. 3D is a growing machine vision application area, especially for the bin-picking challenge, so I would expect to see NET bring out some 3D products in the near future. Now SICK and AT with their structured light products pretty much have the 3D market sewn up, but maybe there’s room for a stereo approach too. If so, NET may be well positioned to be first to market.

A new player in machine vision?

Back on September 23rd I asked if anyone could enlighten me about Ximea. Well if you went to Vision 2010 you will have had an opportunity to learn more, and if you didn’t get to the show there’s an interesting piece about the company in the “Inspect” magazine (Volume 11, Oktober 2010, published by www.inspect-online.com)

Ximea, it turns out, grew out of a vision software company called Softhard Technology. Way back in 2009 Softhard demonstrated a “camera system family” called the Currera and described it as “an intelligent vision appliance…” Well there have been quite a few such products, so I for one didn’t pay a whole of attention at the time.

Now it seems that might have been a mistake. Apparently I missed out on the birth of “… a new class of camera systems [which will] revolutionize machine vision.”

How’s that for hype?

Well I’ve taken a look at the specs for the Currera, and I have to say that it looks pretty interesting. To me it’s a smart camera on steroids – picture the Sony smart cameras in a DVT body and you’ll get the idea. It uses an Intel Atom processor to run Windows Embedded, XP or Linux, and offers “native support for Halcon, Common Vision Blox, etc” (not entirely sure what that means,) plus drivers for use with LabVIEW, MIL and so on. There’s a GigE interface with PoE for getting images and data out, so this is a pretty state-of-the-art piece of hardware.

Imaging is achieved through a 1/1.8” 1.3Mp CMOS sensor from e2v, which I find interesting. e2v are steadily building a reputation as maker of higher-end line scan sensors, so I shall be eager to see just what kind of performance this little beastie delivers.

Pricing starts at €1,666 (which includes German VAT,) and that strikes me as pretty competitive. In fact I’m so curious that I’m starting to think I should go after the North American distribution rights…

Well it’s a thought.

Don’t overlook Matrox

Cognex have done a terrific job of making their InSight smart camera range the tool of choice for any manufacturing engineer needing to implement machine vision. The yellow housings are instantly recognizable and engineers know the software will work. Back in the ‘70’s ‘they’ used to say that no one ever got fired for buying IBM. Well the same goes for Cognex in the machine vision world: it’s the safe choice so why shop around?

Well before handing over your credit card to the folks in Natick (or one of their distributors,) I suggest you do a little “due diligence” and take a look at the Matrox Iris GT smart camera. OK, so a camera is a camera, but the software merits a closer look. Matrox have been developing their “Design Assistant” package to the point where it’s a worthy alternative to InSight Explorer. The latest release, 2.3, (due out any day) improves upon their part fixturing capabilities and adds a bead tool, which I think is quite useful. In addition, and I realize this is just personal choice, I prefer the flowchart approach to programming/development taken by Matrox in Design Assistant to the InSight spreadsheet. Your experience may of course be different.

That’s all for now. I just wanted to say, don’t automatically plump for Cognex. You owe it to yourself (and your employer,) to shop around.

Stereo smart camera?

It’s an interesting idea, (check out the VCSBC64XX model at the bottom of the page of the Vision Components web site,) but is the machine vision world ready for it?

Presumably the target applications are those that need 3D, such as robot guidance, but I have to wonder if stereo vision is really ready for prime-time. Most users of machine vision are still struggling to get their 2D systems to run reliably, so perhaps this is a little too “bleeding edge.”

There again, perhaps a stereo smart camera is just what the market needs to demystify the application of 3D.

No info on pricing but a glance at the spec sheet shows resolutions from VGA up to 3Mp and 240 fps, so it seems like a handy little beastie.

You can always rely on Vision Components to try something new.

More on next-gen smart cameras

A flyer from Vision Components popped up in my in-box the other day to tell me about their 5Mp smart camera, the snappily-named, VC4012nano. This little baby, no bigger than a deck of cards, is available for the princely sum of $999 (€699) which in terms of pixels per dollar, makes it an absolute winner.

Vision Components have long been at the leading edge when it comes to smart cameras, but I’ve always had a problem with their preference for using processors from Texas Instruments. I’m no embedded systems IT geek, but as I understand these things it’s hard if not impossible to get regular Wintel packages to run on such hardware. This means that unless you’re willing to climb the learning curve associated with VC’s own software, they’re not a practical option for the machine vision end user.

However, German compatriot Eye Vision Technology, have taken the VC4012nano into their fold and equipped it with their EyeSpector software. In this more user-friendly guise it sells for &1,500 as the EyeSpector 800.

I’ve not been “hands-on” with EyeSpector, so if any of you readers out there would like to share your experience, please use the Comment function to tell us what you think.

Next generation smart camera

Am I being a little slow in catching on to the new wave in smart cameras? Perhaps I am, because it’s taken Slovakian camera-maker Softhard to make me realize that smart cameras are now software-independent.

This revelation was prompted by news of their Currera range, built around the Intel Atom processor and running Windows XP. Unlike earlier generations of smart cameras (DVT, Banner, Cognex InSight, PPT and so on,) that were wedded to their makers own software, an XP base means I can load up any software I want.

This means, I think, that I could load Sherlock on one camera, VisionPro on a second and Halcon on a third. I realize I might not see a whole lot of speed out of these camera-PC hybrids, but not every application needs intensive processing capability.

I also realize that Softhard aren’t the first to market with this kind of software-independed smart camera. Sony have had a similar product out in the market for some time, and you might argue that Vision Components were there first, although theirs was not a “Wintel” approach.

The PC-smart camera; it’s a new paradigm in machine vision technology and I suspect it might be the future.

Dalsa gets smart(er)

To me Dalsa is first and foremost a camera company. If I need a line scan camera, they are the people I turn to. But that’s understating their capabilities by a long way. They can provide software, (Sherlock, Sapera,) frame grabbers, and a line of “Vision Appliances” from a division they christened iPd.

The iPd line had me a little puzzled from the start. I admired their desire to make vision easier to use, but providing reduced functionality didn’t seem the way to do it. However, with the BOA, Dalsa seem to be taking a different approach to the smart camera business.

There’s a good write up of the BOA in the October 2009 edition of Control Engineering, (Single Point Machine Vision,) so I’ll spare you the details. The key point though is that this is an InSight Micro-sized smart camera that packs the iNspect software from the iPd line. The pricing, at $1,995 for the VGA resolution, should make the BOA worthy of consideration.

So far there are no higher resolution models, although I have to believe they’re in development. According to Control Engineering, Dalsa are also planning to put Sherlock on the BOA. Now when then happens I really will be interested!