Wednesday, April 24, 2013

The eyes have it

Camera-based surveillance systems have definitely played an important role in helping to keep crime down. With the knowledge that their activities will be captured on cameras, members of the criminal fraternity have been dissuaded from committing felonious acts in the community.

But while such systems are undoubtedly effective, they do cost money to commission and maintain. And that's cash that many hard-up communities may be loathed to part with in these financially challenging times.

So could there be a cheaper way to reduce crime without the use of such cameras? Well, apparently, yes, there is. Researchers at Newcastle University (Newcastle, UK) have now discovered that bicycle theft, for example, can be significantly reduced simply by placing pictures of staring eyes above bike racks.

In a two year experiment on the university campus, the academics showed that the eye  pictures -- which were combined with a short anti-theft message -- reduced thefts from the bike racks by 62 per cent.

Newcastle University's Professor Daniel Nettle said that the images of eyes could act to dissuade crime by making people feel that they are being observed -- in a similar way to surveillance cameras -- and as a result behave in a more honest fashion.

That, of course, is the good news. The bad news is that there was also a noticeable difference in crime in places without the signs, where bike theft went up by 63 per cent, suggesting that the crime had been displaced to other locations, rather than eliminated.

Despite that fact, the British Transport Police are now trialing the idea with train Company C2C on a route between Fenchurch Street Station in London and Southend in Essex.

While the idea undoubtedly has its merits, I'd like to think that a more comprehensive solution to the bike theft issue might be to install a couple of surveillance cameras behind the pictures of the staring eyes.

Although my belt-and-braces idea might cost a few more shillings to implement, the combination of the eye pictures and the vision-based solution would not only lead to an even greater reduction in bicycle thefts, but also provide the police with detailed images of those still intent on a life of crime.

Banking on a barcode

If you are a private investor engaged in online trading and banking, having a Trojan attack your PC and whisk your personal financial details off into the nether regions of the internet is a rather horrid experience.

Fortunately, some rather clever chaps at Cambridge University spin-out Cronto (Cambridge, UK) have now developed a system called CrontoSign to address this issue -- a data security solution that makes use of nothing less than a two-dimensional barcode.

In use, a bank generates the proprietary two-dimensional barcode -- a matrix of colored dots containing a cryptographically-encoded message -- and then sends it to a customer. The code is then decoded by the customer using an app running on a handheld device such as a cell phone or on dedicated hardware supplied by the company.

The bar code provides a secure "envelope" around the data so that it can be displayed to the customer over any unsecured channel. So although a Trojan might see the image being sent by the bank, it cannot change the secure data inside.

Two German banks -- comdirect bank and Commerzbank -- have already rolled out the system, which is known in Germany as photoTAN.

Customers can now scan a photoTAN image displayed on the banks' websites using the photoTAN mobile app or dedicated photoTAN hardware device. A customer then sees the message from their bank, which typically asks them to confirm the action they are attempting to perform.

To confirm the transaction, the customer uses a six-digit code, generated by the app or device, and enters it into the browser on their PC. The code acts as the customer's signature for a specific instruction, and once received and validated by the bank's server, completes the transaction.

While Cronto is currently focused on the online banking sector, the company also sees commercial possibilities for the system in e-commerce, peer-to-peer online payments, or any other application where there is a need to create a trusted connection between two parties.

You can read more about the CrontoSign system here. A video demonstrating the system is also available on YouTube in German here.

Thursday, April 18, 2013

Crash detector

Using one of those new-fangled computer tablets while walking along the street can be a dangerous affair.

Just the other day, for example, I saw one self-absorbed individual who collided with another pedestrian while strolling down a pedestrian precinct as he used such a tablet to surf the internet.

It could have been a whole lot worse. He could have walked into something a lot harder, such as a brick wall or a lamp post, and caused some serious injuries to either himself or the infrastructure.

One answer to this problem, of course, is not to use such mobile devices while walking, and concentrate on negotiating the environment instead. But these days, when we all like to be permanently wired into the web, many individuals are unlikely to heed such practical advice.

Recognizing that fact, a team of researchers at the University of Manitoba (Winnipeg, Manitoba, Canada) have now developed a rather nifty little vision-based system that could be the answer to mobile users' prayers.

Their so-called "CrashAlert" system augments mobile devices such as tablets with a Microsoft Kinect depth camera to provide distance and location visual cues of obstacles on a user's path. The Kinect camera itself is connected to a battery powered laptop computer carried in a backpack via a USB connection. When it receives images from the Kinect, it processes them and sends them off to the tablet via a Bluetooth connection.

In this way, a user can see their surroundings on the tablet while they walk, dodging and slowing down or lifting their head to avoid any potential collisions and related injuries.

Now the cynics amongst my blog followers might consider that hauling around a bulky computer and a Kinect system in a backpack completely defeats the purpose of using a lightweight tablet in the first place. And, of course, they're probably right.

But if such a system was miniaturized and actually fitted to the tablet itself, then it might actually prove to be of some practical use. And I'm sure that such systems will be in the future.

A research paper entitled "Crash Alert: Enhancing Peripheral Alertness for Eyes-Busy Mobile Interaction while Walking," by Manitoba University researchers Juan David HincapiƩ-Ramos and Pourang Irani is available on the internet here. Just don't read it on a tablet while attempting to negotiate a busy pedestrian precinct.

Editor's note: Interested in reading more about novel uses of the Kinect? Then why not browse through our recent slideshow here.

Tuesday, April 16, 2013

Reading and running

Ask anyone and they will tell you the same thing. I've never been too fond of outdoor sports. My dislike probably dates back to my time at school in England, where all the lads were required to take part in rather rough games of "rugger"” three times a week.

Nevertheless, I can certainly see the advantage of giving the heart and muscles a good old work out in the comfort of my own home on one of those new fangled running machines.

But the problem with those running machines is that, up until now, it’s been impossible to partake in anything more intellectual -- like reading a treatise on how to program computer vision systems with Python -- whilst pounding away on the treadmill.

Thankfully though, some rather clever chaps at Purdue University (West Lafayette, IN, USA) have now come up with a solution to this problem. That's right. They have developed a system called "Readmate" that allows treadmill users to read text on a small monitor mounted in front of the machine while they are exercising.

To do so, a user must first don a pair of goggles equipped with infrared LEDs. An infrared camera can then track the runner's bobbing head by capturing images from the LEDs. Then the text on the screen is moved in unison with the head movement.

According to Ji Soo Yi, an assistant professor of industrial engineering at Purdue, the text cannot be moved exactly in synch with the head because the eye is already doing what it can to compensate. So the system accounts for that compensation by moving the text slightly out of synch with the head motion.

While the new system could prove to be a boon for those who get easily bored by endlessly running on the same spot, it also might be used by heavy equipment operators and aircraft pilots who experience heavy shaking and turbulence while reading information from a display.

Thursday, April 11, 2013

Heavy goods

If heavy goods vehicles and their trailers are too heavily loaded, or the loads incorrectly distributed, they may constitute a traffic hazard and damage road surfaces.

But stopping vehicles randomly at weigh stations for no good reason -- especially those weighing hundreds of tons – can mean that a lot of fuel is wasted unnecessarily in the stopping and starting process.

Now though, SINTEF, the largest independent research organization in Scandinavia, is heading up the development of a new system called "NonStop" that could offer a novel solution to the weighty problem.

The system makes use of a special piezoelectric cable countersunk into the road surface. The cable generates an electrical voltage when subjected to pressure, and in this way the weight of a vehicle passing over it can be determined and recorded by a computer.

Complementing the road sensor is an automatic number plate recognition system that will read vehicle number plates from which a vehicle's permitted load can then be determined.

The measured weight and the load that the vehicle is allowed to carry will then be used by inspectors from the Norwegian Public Roads Administration (NPRA) to assess whether it should be stopped or not.

SINTEF was commissioned by the NPRA to develop the system. Other partners involved are the Norwegian Hauliers' Association and the Oslo firm Ciber.

What I particularly like about the Scandinavian idea is that it not only makes use of state of the art vision systems, but uses them in conjunction with a cable whose piezoelectric properties were discovered way back in 1880 by French physicists Jacques and Pierre Curie.

Indeed, considering how to couple older technologies with the new might also provide many developers of vision systems in other fields some ideas along similar lines.

Wednesday, April 10, 2013

Magic mirror

Regular readers to this blog will be aware of the coverage that I have devoted in the past to a company called Raspberry Pi,  a spin out from that UK hive of cerebral activity, Cambridge University (Cambridge, UK).

The engineers at Raspberry Pi have developed, and are now selling, a small inexpensive Arm (Cambridge, UK) based computer that plugs into a TV and a keyboard.  It's a capable little PC which can be used for many of the things that a desktop PC does, like spreadsheets, word-processing and games. It also plays high-definition video.

Since its launch, the inexpensive computer has attracted a lot of interest from hobbyists and academics alike who have deployed it in a variety of innovative ways.

In one of the more recent applications, a French chap by the name of Pierre Raufast has used his Raspberry Pi computer, a webcam and OpenCV software to create a "Magic Mirror" with a disembodied voice which recognizes the person looking into it and responds accordingly.

To enable others to build a similar system, the generous Frenchman has posted up an easy-to-follow tutorial, including a hardware list, software, instructions and tips on successfully using OpenCV for face-recognition. It can be found on the Think RPI web site here.

Having finished assembling the hardware, downloading and compiling the source code and training your system to recognize individuals, Monsieur Raufast recommends that you take a break and read "L'homme qui plantait des arbres," an allegorical tale by French author Jean Giono, published in 1953.

Personally, if I were Monsieur Raufast, I wouldn't be sitting back on my laurels and reading anything. If I had developed such a system, I'd be investigating whether the good folks at The Walt Disney Company might be interested in parting with some of their money to help me commercialize it.

Monsieur Raufast has posted a video of his Magic Mirror in action on YouTube. You can view it here. I wonder what the Brothers Grimm would make of it?

Thursday, April 4, 2013

Teddy cam

A camera hidden in a teddy bear has caught a care worker stealing money from the home of an old age pensioner in the UK.

According to a report in the UK's Daily Telegraph, 28-year old care worker Emelie Kleen-Barry was caught red handed stealing from 81-year old grandmother Margret Birch by the hidden camera. She was jailed at Leicester Crown Court for 13 months for stealing £40 in cash from Birch's home.

Mrs Birch kept her purse in a wardrobe in her room, but when funds appeared to be diminishing faster than she expected, her family planted the so-called "Teddy Cam" in her room and focused it on the wardrobe to find out what was happening to the money.

The family called the trap 'Operation Narnia' and the device soon came into its own when it caught Kleen-Barry taking money from the grandmother's purse.

"In my professional opinion, Teddy Cams and other hidden cameras are the best way to give peace of mind to families and friends who are worried about what goes on when they can't be around to care for vulnerable people," says Kristy George, a spokesperson for the Birmingham, UK-based firm Private Detective (Birmingham, UK).

That company claims to have had great success in the past with Teddy Cams. Not only are they an excellent way to keep an eye on an elderly or infirm relative of friend who lives alone or in warden controlled home, they can also can be given to a child to find out how they are being cared for by others when the parents are not around.

Clearly, such wireless monitoring devices can be a great boon for those with elderly parents or those with small children. But such Teddy Cameras are not inexpensive. One such bear from Eyetek Surveillance (Chaddesden, UK) that comes equipped with a color camera and transmitter which transmits video to a receiver currently retails in the UK for £145 (almost $220).

Faced with similar surveillance issues, those of us with an engineering bent might like to consider constructing our own such bears instead of buying one. Certainly, it's an idea that has caught the imagination of our European Editor Dave Wilson.

He tells me that he could pick up a wireless camera and receiver for around £45 ($68) from UK electronics gizmo supplier Maplin that he is almost certain he could retrofit quite easily to the somewhat elderly bear from his childhood.

Soon then, the unfortunate beast might be taken from the upstairs cupboard where he has lived for the past fifty years only to have his soft delicate insides drawn out to make way for the new vision-based implant.


1. Thieving care worker caught on camera inside teddy bear

2. Eyetek Surveillance Wireless Teddy Camera 

3. Discreet Wireless Colour CCTV Camera

Zero defects

Many reasons are often cited for deploying machine vision systems. These include improving tedious and repetitive manual inspection tasks that are prone to human error while increasing productivity at the same time.

In many manufacturing environments, single point inspection systems can determine whether a product has been properly assembled. If not, the product is then rejected and may be reworked or scrapped. Needless to say, such reworking may prove expensive if a product has gone through multiple manufacturing stages before being inspected.

To reduce the amount of reworking required, many manufacturers employ what is known as a "zero defects forward" approach. Rather than inspect a product after it has been fully assembled, the product is inspected after each step in the assembly process.

In this way, any defects that occur at each stage are recognized and can be more easily corrected before the next phase of assembly. In addition, such "zero defects forward" approaches save the time and money that would have otherwise be wasted by completing the assembly of a defective product that would ultimately need to be disassembled and then re-assembled.

To further improve the manufacturing process, manufacturers can also deploy vision-based robotics systems to automate the assembly task itself, relieving operators of the tedious job of doing so.

By deploying such systems in conjunction with vision-based inspection systems, manufacturers can further reduce assembly cost, increase productivity and eliminate human error. However, for many Smaller to Medium sized Enterprises (SMEs), implementing fully automated robotic assembly systems may only be justified if the return on investment is high enough.

As an alternative, those enterprises who have already realized the benefits of implementing vision systems to inspect their products could consider deploy semi-automated assembly systems to evaluate their effectiveness.

In doing so, they will avoid the costs of implementing fully automated assembly systems, while at the same time reaping the benefits of semi-automated assembly before they ultimately and inexorably move towards a totally automated manufacturing environment.