How CoaXPress and Machine Vision Are Giving Traffic Monitoring a Fast Commute to the Future

As the world develops more efficient methods of transportation on our roads and highways, ways of managing the influx in vehicles must keep pace.

(Courtesy BitFlow)

From the prevention of red light runners and toll evaders to daily traffic reports via the web and traffic counting evaluations for future road expansions, machine vision is fast becoming the means to safer, more coordinated, and more “intelligent” use of our roads. The late 1970s gave birth to the first Intelligent Traffic System (ITS) in the form of the license plate reader. Since then, several companies have developed machine vision technology for ITS to a point where efficiency and accuracy are at their highest levels ever.

Analog data in these systems is transmitted at a sluggish 11Mbps over a single coax cable from the camera to the computers. Now, consider the potential volume of data over a single coax cable with just one CXP link at 650Mbps—we’re talking almost 60 times more data.

Machine Vision and Intelligent Traffic Systems
Essentially, machine vision is the ability to use a camera, computer, and an illumination source to make a decision. For all its sophistication, an Automatic License Plate Recognition (ALPR) system is basically comprised of the same components as an imaging solution for checking the quality of toothbrush bristles, for example. Technological developments have changed over the years, such as lighting taking the infrared illumination path and then LED, and options such as strobing and flash have emerged. Yet the basic components remain the same: Camera, illumination, and PC.

Traffic Analysis is another application under the ITS umbrella where imaging is deployed. Instead of concentrating on license plates, it inspects traffic itself. High-speed cameras frame the roadway to detect the volume of free-flowing traffic, as well as the speed at which it is moving through the frame. In areas of high traffic or where a traffic study is needed to determine the need for extra lanes, Traffic Analysis is a helpful tool. It is also valuable as the live video feed can be given to news networks and law enforcement or streamed on the Internet, showing us what the traffic looks like in real-time at a specific location.

Figure 1: The Aon-CXP frame grabber is part of a trend that is seeing CoaXPress cameras shrinking while also offering higher performance and improved power efficiency. (Courtesy BitFlow, Inc.)

Advancements in computer chips have solved many ITS problems, with systems getting cheaper and faster every three to six months. PCs are now fast enough to allow database matching for blacklist and whitelist applications in real-time, for instance, to prevent unauthorized entry into buildings. Cameras, once analog across the traffic world, have migrated to digital with the option now for CMOS over CCD cameras. Over the last number of years, CMOS has emerged to be lower cost, with lower noise issues in visible light and requiring lower power to run. All these are indicative of CMOS cameras being the better option for the majority of ITS applications. While typically in a NEMA protective housing, cameras for traffic applications are more robust than ever before, able to handle temperatures ranging from -30°C to 75°C (-22°F to 170°F) and 10%-90% condensing humidity. Unlike in machine vision applications, cameras in traffic systems need to manage variable lighting conditions, unpredictable appearance of cars, long distances to the processing system, and changing weather conditions. Consequently, low-end machine vision cameras are not suitable for most of these systems. Instead, they need to rely on high-performance cameras to increase accuracy—regardless of traffic volume, speed, or weather conditions.

Machine Vision Interfaces
Besides camera, illumination, and PC, another aspect in system design is the machine vision interface between the camera and the computer. Cameras need to transmit bandwidth-heavy data to the PC so that data is not lost and arrives without lag over long distances. When analog cameras were in use, and some of them still are, the interface was a BNC coaxial connection between the camera and a data acquisition board, known as a frame grabber. The frame grabber was usually a PCI interface, fitted into the motherboard, and allowed data to transfer seamlessly between the analog camera and computer memory.

As technology’s march has continued forward, so has frame grabber design. Now enabled are digital interfaces such as LVDS, RS422, Camera Link, and most recently, CoaXPress (CXP). With the advent of these additional interfaces, camera choice has broadened, encompassing, for example, camera sensor sizes, options such as IR/monochrome or color, and frame rates. Of these interfaces, CXP appears to be the most promising for ITS.

The Japan Industrial Imaging Association formally introduced the CoaXPress V1.0 high-speed point to point serial communication standard back in 2010, followed by CXP V1.1 in 2013. Designed for transmission over single or multiple coaxial cables, the original version had a high-speed downlink of up to 6.25Gbps per cable, plus a lower speed, 20Mbps uplink for communications and control. Power is also available over the cable (“Power-over-Coax”), and cable lengths of greater than 130m may be achieved without an extender. CoaXPress V2.0 is due to be released in mid 2018 and will add two speeds: 10 Gbps (CXP-10) and 12.5 Gbps (CXP-12). Uplink speed will also be doubled to 40 Mbps for CXP-10 and CXP-12, allowing trigger rates over 500kHz without requiring a dedicated high-speed uplink cable.

CoaXPress has been one of the fastest growing interfaces in the machine vision industry over the past  three to four years. Because of the low cost of coaxial cables, and its ultra-fast data transmission rates and achievable distances, CXP is certain to remain a major player in the coming years.

Sluggish to Speedy
With signal latency a critical consideration for vehicle tracking and highway control applications, the CoaXPress standard can be useful, as it allows the transfer of uncompressed video, making zero latency possible. High frequency real-time triggering and exposure time adjustment can also be accommodated.

To get a better sense of CXP capability, let’s look again at (Automatic) License Plate Readers. In Europe, the standard license plate is easy enough for a machine vision system to read. The plates are 18 x 20 inches. Letters/numbers appear four inches high in large, bold unhindered fonts. The story in the United States is different, as the plate is 12 x 6 inches. Apart from up to seven alpha-numeric characters, there is also a state name and sometimes an image superimposed on the plate. This can lead to confusion with a low-resolution image. Yet using a high-resolution camera demands an advanced interface to transmit the data back to the processing center quickly and accurately. This is where the advantages of a CoaXPress interface come into play.

For example, consider a toll plaza where a car drives through without paying. Current technology can capture the image, but making the distinction between the plate being recognized as “ABC123” or “A8CIZ3” requires the processing center to verify the car make and interaction with the registry. This, in turn, dramatically reduces efficiency of what should be a simple read if the resolution and data delivery were fast enough. In traffic analysis the requirement for CoaXPress is even more pronounced. Currently there are thousands of analog traffic cameras in place on our highways. Analog data in these systems is transmitted at a sluggish 11Mbps over coax cable from the camera to the computers.

Now, consider the potential volume of data over a single coax cable with just one CXP link at 650Mbps—we’re talking almost 60 times more data. This would enable images of traffic on routes to be crystal clear, and also allow for better dissemination of information in terms of traffic backlogs, accidents and break downs, thus improving overall driver safety.

This takes us to the chief advantage of CXP in traffic applications, besides its high speed: retrofitting. CoaXPress uses the same coaxial cable network as the analog cameras, making retrofits cheaper and easier. Swapout of a board and camera is all that’s needed to obtain high-quality, high-speed CXP images. In fact, these enhanced images are clear enough to easily see license plate numbers on a smartphone app. And because CXP offers Power-over-Coax, there is no need for additional cabling work. Power over CoaXPress applies a constant 24V DC voltage to the cable core, without influencing the signal integrity, to provide up to 13W per cable. In short, CoaXPress combines the simplicity of off-the-shelf 75-ohm coaxial cable with high-speed serial data technology.

In conclusion, CoaXPress can play a vital role in the many applications that make up ITS, two of which were touched upon in this article. As designers sit down with this interface at their disposal, more ideas will be formed, and CoaXPress will take its place among the important developments to enable ITS to perform more efficiently.

Donal Waide is the Director of Sales for BitFlow, Woburn, MA USA. Waide can be reached at or +1-781-932-2900. Donal began his life in machine vision in the ITS sector in the 1990’s and has been with BitFlow since 2010.

Share and Enjoy:
  • Digg
  • Sphinn
  • Facebook
  • Mixx
  • Google
  • TwitThis