Through Different Eyes - 25 Years of MATRIX VISION

A medium-sized company near Stuttgart has become a driving force in the world of image processing. Existing applications represent the very latest state of the art.

What many of us consider an inconvenience is often rendered a sheer delight by our children's fascination. And the more technically minded know that it is all down to high-tech equipment. Watching the machine take up the deposit bottle, rapidly rotating it around its own axis and swallowing it whole, we are all grateful that the bottle returning process is no longer a laborious manual exercise, where every bottle has to be individually inspected and the deposit value entered in the till.

The convenience offered by the cameras in these machines may now be taken for granted, but not so long ago it was very much the stuff of visionaries – like those working for a company in Oppenweiler in Germany's Rems-Murr district. This small town of 4000 inhabitants in the Swabian Forest is where Matrix Vision GmbH has its headquarters. The company was founded by two technological trendsetters, Gerhard Thullner and Werner Armingeon, 25 years ago and soon made a name for itself with its clever new ideas.

The reverse vending machines that are now found at many discount stores are just one example of this ingenuity. Uwe Furtner, junior partner at Matrix Vision, made a significant contribution to the success of this extensive project. And the demands of both Deutsche Pfandsystem GmbH (DPG) and the German retail sector were high. All suppliers had to be DPG-certified.

Not only did the reader unit have to reliably detect the DPG security mark, but it also had to be accommodated within the confined space of the machine. To meet this challenge, the Matrix Vision developers constructed a Security Mark Reader (mvSMR). The company sees two main advantages in this intelligent camera system: Firstly, the captured images can be processed directly in the camera unit and, secondly, the camera can control the various exposure sequences of the lighting units in order to detect the deposit mark.

To process the information, the camera unit is equipped with a 400 MHz processor and a 1600 x 1200 pixel sensor. This sensor is a necessary requirement since bottles of varying sizes can be inserted bidirectionally. The camera must therefore cover a sufficiently large field of vision. The mvSMR supports bottle diameters up to 120 mm and a rotation speed of up to 25 m/min. The robust and reliable detection algorithm can accurately identify the security mark even on slightly deformed or dirty surfaces.

The following processes are triggered via the digital outputs: When the mark is detected, the deposit is credited and the bottle transported away. The bottles are then destroyed to prevent multiple deposit returns. Such integrated solutions not only save space, but they also negate the need for other components, such as barcode readers, and thus help to reduce costs.

One particular aspect is critical to many image processing applications: light. To provide the artificial eye with the necessary field of vision, the detection environment must be well illuminated. This applies particularly to deposit bottles, which are often a little dirty and can have minor denting, and the detection of which is greatly affected by shadows. Since none of the conventional lighting components offered satisfactory results, Matrix Vision simply developed its own lighting unit.

The company has also made contingencies for the future: If the deposit system should ever have to meet other or additional requirements, the mvSMR software can be upgraded online accordingly. Any legal amendments or new DPG specifications can thus be quickly and easily implemented.

The entire image processing unit of the Oppenweiler company was not only well received by the DPG study group supporting the introduction of the reverse vending machines. Discount stores were also keen to employ this solution comprehensively. However, to meet the tight deadline for the coming into force of the new deposit legislation, Matrix Vision had to overcome yet another challenge: In a period of just three months, the medium-sized company delivered no less than 10,000 systems.

It is no coincidence that the Swabian image processing specialists are constantly entrusted with such demanding assignments. The company, which celebrates its 25th anniversary in 2011, has enjoyed a history full of technical advancements. Along with Furtner, the two founders, Armingeon and Thullner, have been like lighthouse builders in the agitated sea of data processing, continuously showing the branch of trade the technical landmarks.

The intelligent cameras are just one example. They have now become indispensable to many thousands of applications and sectors in the fields of quality assurance, medicine, surveillance, and traffic control. However, these digital eyes with integrated computers were virtually unheard of just a few years ago. Matrix Vision can rightly claim to have launched one of the first truly intelligent cameras. Time and again, Armingeon had returned to this concept. "If I had been able to devote myself more fully to this idea, we could have brought the product to market much sooner," says the development manager with hindsight.

The main aspect of these systems, which are also known as smart cameras, is that they not only record images but can also interpret these themselves. They are effectively equipped with their own PCs, which can extract application-specific information from captured images before processing it. In a quality assurance application, for example, the camera can decide itself whether a part is good or bad.

The fact that Armingeon thought to combine cameras and intelligent systems in 1999 is no great wonder considering the two engineers’ significant technological contributions, many of which are an integral part of today's image processing solutions. What is most impressive is that the pair successfully navigated the pitfalls associated with ever-advancing technology as well as those posed by the general zeitgeist and modern marketing.

It is a well known fact that numerous major players and inventions in the data processing industry remained oblivious or unaffected by trends in technology and marketing. One particular example is Atari. In the early years of computer systems, this company's products were in great demand with early adopters and universities. As one of their first business ventures, Thullner and Armingeon developed a graphic controller for a 19-inch screen for Atari computers – mainly because the company didn't care to do so itself, but also because Mac and PC users were already looking at the bigger picture. After just six weeks of development, they introduced the innovative new product at the CeBIT trade fair in Hanover in 1989.

With this world first, the pair had demonstrated their successful cooperation, with Thullner's market sense and Armingeon's inventive spirit standing them in good stead to this very day. Looking back on this period, Thullner states that they had achieved "excellent sales as a global market leader".

When Atari was left behind by the rise of PCs, the two engineers had already moved on to pastures new. They had briefly considered applying their knowledge and skills to the booming PC market but soon thought otherwise. "There were already established companies, and we wouldn't have been able to compete," says Thullner in explaining their decision.

From this point on, the team dedicated itself to the development of frame grabbers for industrial applications. These electronic components are used to digitize analog video signals. With the image processing sector still in its infancy, Armingeon and Thullner successfully made their mark with these products, which can still be found in the company's range today. However, the digitization of camera systems has reduced the significance of this once revolutionary technology. "The technology has moved into the camera," explains Armingeon.

The image processing market is also undergoing constant change, something which Armingeon, Thullner, and Furtner keep a very close eye on. They are all too familiar with the pitfalls of the mass market and are thus keen to avoid it. They have adopted a different approach: if it is beneficial to the customer, components from the mass market are also employed in industrial solutions.

"We are driven by technology and frequently adopt a pioneering role in new developments," is how Armingeon summarizes the experience gained throughout his company's 25-year history. "However, once a product starts to be manufactured in large volumes, our dedication as a development company stops." As Thullner puts it: "We are always riding the peak of the technological wave."

None of this would be possible without a well-established team of specialists to effectively support the company's successful balancing of continuity and change. "We can rely on our exceptional workforce to promote change and advancement in the world of image processing, while also safeguarding the future of tried and tested solutions," says the proud company director.

Matrix Vision is currently accompanying its customers through a period of transition in the field of sensors. The two technologies, CCD and CMOS, are currently available on the market. The newer CMOS sensors allow more cost-effective manufacturing of products and also enable initial signal processing directly on the chip. However, the newer solution is not always the ideal one. Furtner explains: "The choice of sensor is very much application-specific." Armingeon adds that CCD sensors are generally characterized by better image quality, lower noise, and an absence of pixel defects.

"The choice between a color or grayscale sensor usually depends on the task to be performed," is Matrix Vision's recommendation. Some sensors are only available in one format or the other. Color sensors feature a color filter structure in front of the light-sensitive sensor matrix, as a result of which specific pixels only absorb light of a specific color.

"These filter structures are permeable to infrared light. To avoid inaccurate color representation, an additional infrared blocking filter is required. However, due to the pixel-based color change, use of this filter results in a lower spatial resolution when recording colored objects," Armingeon continues. In the case of applications requiring high color accuracy, such as the color check of printouts, or high spatial color resolution, the use of 3-chip cameras is recommended, as these employ separate chips for the colors red, green, and blue. "Another important aspect is the shutter," states Armingeon. "CCD and CMOS sensors can feature global shutters, while simple CMOS sensors generally have a rolling shutter. In the latter case, capturing fast moving objects often results in geometric distortion."

All these technical intricacies show that expertise is paramount to the selection of image processing components. "We always develop solutions in close cooperation with the customer," says Furtner. "This ensures that the right technology is employed in all cases."

And in some cases, the company is required to develop appropriate technology for the customer – but this is no problem for Matrix Vision GmbH. After all, it is this very spirit that has characterized the company's 25-year history.

Company history: change as a constant

  • 1986 Company founded by Werner Armingeon and Gerhard Thullner
  • 1988 World's first high-res graphics card for ATARI
  • 1992 Transition from ATARI to industrial image processing
  • 1992-93 ISA bus frame grabbers PCgrab-G1/G2, PCprocess
  • 1995 PCI bus frame grabber PCimage series, Uwe Furtner becomes partner
  • 1999 Intelligent camera mvCAM
  • 2001 Start of development of mvIMPACT image processing software
  • 2002 Embedded Linux camera mvBLueLYNX
  • 2003 Separate offices/representation in France and Italy
  • 2004 Industrial USB2.0 camera mvBIueFOX
  • 2006 mvHYPERION frame grabber series for PCI Express®
    • Industrial GigE camera mvBIueCOUGAR
    • Products for bottle deposit system
    • DIN EN ISO 9001 certification
  • 2008 IBM Cell components
  • 2010 Next X-generations of GigE and intelligent cameras

Go back