On the face of it, easy. But on closer inspection, challenging.

That’s oil drum internal inspection

When a customer asked us to provide a camera that would allow them to inspect the insides of new oil drums at the end of the production line, we made the assumption that any reasonably small camera would be good enough for the task.

However, the standard oil drum has a 2″ aperture at the top through which any camera assembly would need to fit. There are many small cameras on the market that would fit through a 2″ aperture. But taking into account the operational requirements that meant attaching the camera to an existing probe that also has to fit through the same space, meant that the majority of small industrial cameras are too big.


In addition, the specification was made more challenging because:

  • Every centimetre of the internals of the drum had to be inspected, requiring more than one camera
  • The cameras had to enter the drum and exit within 30 seconds
  • The live images needed to be displayed on a large screen
  • High resolution images were required so that the smallest objects could be detected
  • Allowing for routing around and through a moving production line, the cable distance was around 15 metres
  • The whole process had to be semi-automated so that the operators only have to watch the screen

There is one camera in our portfolio which stood out as the best candidate. Cable run had to be minimal and so power over ethernet was an obvious choice, meaning the miniature power over ethernet cameras from The Imaging Source were an ideal choice.


The cameras mounted at 180 degrees, with LED strip lights and M12 miniature lenses

Then the next problem became apparent, standard lenses are bigger than the width of the body of the camera, so we had to go to small lenses. Fortunately there is an M12 lens adapter available from The Imaging Source that converts the C mount to an M12 mount. Now we just had to find fish eye lenses with a wide enough angle to view each end of the drum.

Finally, how to illuminate. Standard lighting couldn’t be used as there was nowhere to put it. We came up with the idea of using LED strips which work off 12V DC. These attach to the side of the camera and as they are low power they don’t get very warm.

Once we’d cracked the camera and lighting problem, we then had to display the live images on the big screen. For this we used Scorpion Image Sentinel, designed for uncompressed image capture at live video speeds.

What the inside of an oil drum looks like!


New Scorpion Vision Starter Kits

Read more at Scorpion Vision Web – http://scorpionvision.co.uk/announcements/scorpion-vision-starter-kits

Announcing the Unibrain Scorpion 2D Machine Vision Starter Kit

We’ve launched a new product intended for use by those who are curious about machine vision and believe they may be able to use such a system in their manufacturing environment, but don’t have the confidence yet to invest anything significant into such a venture.

What does it cost to employ one person to check a product is the right colour or size as it passes along a production line? In the UK, that’s an easy number to come up with – without taking into account the additional costs with recruitment, on-site training and induction, the National Minimum Wage from October 2010 will be £5.93 for workers aged 21 or over.

That equates to £12334 for a 5 day week over the year, excluding national insurance payments.

The Unibrain Scorpion Machine Vision Starter Kit

Automation is a way to reduce costs using Scorpion Vision Software. The starter kit which includes everything needed to build a system (just supply a PC) costs £1110 or 1260 Euros. This isn’t a pipe dream. This kind of automation solution is achievable on a budget.

To see more about this, have a look at the relevant product page on our web site: http://scorpionvision.co.uk/catalogue-index/scorpion-vision-products/starter-kits

We even throw in some help for free.

– Paul

Planning for our Technology Day – it’s next week

I visited the venue at Brooklands Museum yesterday to make final arrangements for next week’s event. It’s a great location for a technology

Aerial View of Brooklands circa 1938

Aerial View of Brooklands circa 1938

day as the venue is steeped in engineering history dating back more than a hundred years to when the world’s first motor racing track was built. A fledgling motor industry blossomed into a world class centre of automotive engineering excellence well before the second world war. War brought a change of direction and the site became home to a number of aircraft manufacturers.

Remnants of these bygone times can be seen all around the site including sections of the original banked race track. Some of the cars and aircraft built at Brooklands are on display and browsing these magnificent relics I was prompted to imagine the times when machines were designed on paper and built by hand. There were production lines of course but ‘quality control’ as a specific discipline for which whole departments exist was probably not in the business plan 70 or 80 years ago.

If the engineers of those times could be brought by time machine as special guests for our technology day, what would they have made of our modern manufacturing processes and our highly automated manufacturing systems? What would their reaction be to seeing a smartcamera running Scorpion Vision Software that can measure by itself to micron accuracy?

What would Barnes Wallis or RJ Mitchell think of our 3D vision systems that can identify complex objects in a pile and extract them by sending X,Y Z data to a robot?

The Scorpion Technology Day, Thursday 10th December 2009 at Brooklands Museum, Weybridge, Surrey. 


Scorpion Vision: A Logical Technology for Recycling Systems

recyclingHaving spent two long days last week in a factory making 200 million hooks for retail businesses (and who, with the help of Scorpion want to re-use old, returned product). And this week promoting the use of a system that checks the quality of dough before it enters the oven in a factory producing thousands of bread based products (and diverting those raw products for re-processing) it occurred to me that we aren’t shouting enough about the green credentials that Scorpion Vision Software justly deserves.

This is a part of the modern commercial landscape that we should be laying claim to. It is not by any means a tenuous link, as vision systems like Scorpion can be used to recycle any kind of product or material at high speed – faster and more accurately than humans could ever do.

Incidentally, four years ago I was working with an engineering company to build a cardboard recycling system. Scorpion’s role in eliminatesweakestlinkthis was to monitor waste paper streams in a waste recycling plant. When cardboard is detected on the conveyors, a signal is sent to divert the cardboard away onto a second conveyor to a bailing machine. Previuosly this work was carried out by dozens of low paid immigrant workers. No offence to those workers but they can only do so much with little incentive!

So, the slogon ‘Eliminates Your Weakest Link’ is more appropriate than ever before.

Watch this blog for more details on Scorpion’s role in recycling and other projects that benefit the environment and humanity in general.


Checking For Golf Balls In A Sea of Potatoes

Here’s a simple but novel Scorpion Vision application. Some years ago a well known producer of potato crisps asked if it was possible to detect golf balls

No golfballs detected

No golfballs detected

in the potato conveyor before the potatoes were given the chop. The problem was that the mechanical methods used to remove foreign objects such as metal or stone just didn’t detect the more ‘organic’ make up of the common or garden golf ball, of which there seemed to be a persistent presence in the crop being delivered to the factory.

With Scorpion Vision, this proved to be a simple solution that was deployed by the local machine builder who had very little experience with machine vision systems. With a

Golfballs detected!

Golfballs detected!

little help from us, he configured colour cameras to snapshot 1 square metre of the conveyor, then set up the ColorAnalyser tool to scan each image for the presence of red, white or yellow. The detection of red and white amongst a crop of potatoes is very easy but the yellow proved more challenging as there was a relatively high number of false detections caused by reflections from both potatoes and the rollers on the conveyor. However, these were reduced by utilising the shape and circularity filters in the ColorAnalyser.

The colour analyser tool integrates blob analyses with HSI (Hue Saturation and Intensity) pixel measurements. As blob detection is built into the analyser, standard blob functions are available making this a very powerful and sensitive vision tool.

A complex, multicamera vision system for less than £5k!


Unibrain 70fps Bonsai Camera

Here’s an idea:

Take four Unibrain compact machine vision cameras costing less than £400 each.

Add a Scorpion Lite license for £1070.

Allow say, £800 for a decent industrial PC, another £1000 for lights, cables and power.

And another £300 for optical triggers for the cameras, some aluminium profile for attaching everything to.

Scorpion Vision Software

Scorpion Vision Software

That’s a total of £4770 for the components of a very comprehensive machine vision system that will potentially inspect four production lines simultaneously, many times a second.

But what about development time to create the application? It’s all in the built-in toolbox. All the algortithms are there. A competent person can do it in less than a day depending on the complexities involved.

So, it IS possible to build a vision system to inspect four production lines for less than £5000. Of course there are other costs involved like engineer’s onsite time and training the system with good and bad products for example. But the point is that the real hardware costs can be easily recouped in a very short period of time, demonstrating that the Return On Investment can easily be measured.

Furthermore, with Scorpion Vision much of the programming is already done. The production engineer just has to put it all together using the 150+ vision tool suite at his or her disposal.