Trends in intelligent traffic systems – more machine vision cameras and cameras doing more

Intelligent Transportation Systems (ITS) are constantly becoming more sophisticated and one recent trend is the increased use of machine vision cameras.

We at Adimec have several new customers and many more inquiries to use our cameras in intelligent traffic systems, primarily high speed license plate recognition applications for now. Some other traffic applications for our cameras are discussed here.

Here are some of the trends we have observed:

Move from analog to digital (some digital MV cameras)

Interfaces such as GigE Vision and CoaXPress are enabling this switch as they enable long distances between high data rate digital cameras and the PC.

Replacing multiple cameras with one higher resolution camera

To simplify systems, reduce unnecessary data, and reduce costs, 1 high resolution camera (often HDTV) can replace 2 or more low resolution cameras.  The goal is to have only one camera per lane or two lanes per camera. Current trends in machine vision to introduce higher resolution cameras with 4, 8 or 12 MegaPixel are also followed in traffic.

Replacing SLR cameras with more robust (industrial grade) cameras

It is ideal to have the cameras above the road to minimize skew/obstruction problems that occur from the side of the road.  This means the cameras have to be extremely reliable to avoid closing the road for repairs to the camera.

Triggering with the cameras (rather than induction loops)

Triggering systems in the road are very costly to install and maintain and can have latency issues.  High-speed cameras with the appropriate functionality and combined with software can be used to do this at much lower costs.

Using the camera for multiple purposes

The camera can have multiple uses within one system:

• As a trigger (as mentioned as above)
• Provide an OCR still image (color, high resolution)
• Measure speed by using images from two different distances
• Provide classification of the vehicle as a bus, car, truck, etc. for proper payment per class

Move to CMOS

As the quality of CMOS image sensors are improving, the inter-scene dynamic range, high frame speeds, and low power requirements are making them a very attractive option in traffic systems. One trigger for traffic system manufacturers to consider CMOS is that they inherently do not suffer from blooming and smear problems.

More processing on the camera to minimize demands on the PC

If the cameras are running in high resolution mode all of the time, there is a ton of data and most of it will not be used.  Certain functionality can be added to the camera to reduce wasted data to minimize the overload to the PCs.  One example is for the camera to operate in a region of interest (ROI) mode with just a small portion of the scene.  Once an object is detected in the scene, the camera switches to high resolution mode in order to capture the necessary data.

Conclusion

Cameras, particularly high-performance machine vision cameras, are becoming more important in intelligent traffic systems.  There are some special considerations for utilizing these cameras to these demanding applications which we will expand upon in our blogs this month.