Things to consider when specifying an ALPR camera

License plate recognition for high security sites or municipal/intelligent transportation applications is becoming more popular.  Thus there are more and more products offering LPR. There are 2 main ways to implement an LPR system:

  • Standard IP cameras and LPR software housed on a separate server
  • Specialty cameras specifically designed to perform LPR.

When Genetec first started in the LPR business, we tried our hand with regular cameras and software-based LPR. Based on our experience however, we decided to take the route of a specialized single purpose camera for license plate recognition. Let me explain why. 

License plates are never in the same spot on a car and no one drives in exactly the same pattern. So, one of the challenges for LPR is maximizing field of view to capture as many licenses plates as possible with a single camera. The AutoVu Sharp is built with an high resolution infrared camera where the shutter is synched with the on board illuminator.  As well, there is a second VGA camera built into the same unit to capture an overall "context" image of the vehicle. If you are just using a regular camera, you may be sacrificing plate reads to get a context image.  Or, if you are using an illuminated camera, you may be losing the context image which can be very important in helping to  identify  not only license plates, but the make and model of a vehicle.  In addition, our LPR processing is done entirely on the camera on a dedicated processor. There are a few key advantages to this architecture.

  1. There is no concern about taxing your VMS server by having to run LPR algorithms on the host. Our systems are designed to do the heavy lifting with enough processing power to process dozens of images per second that scale!
  2. Onboard processing also provides the ability to do LPR over very low bandwidth links, as it is not required to transmit the video to do the read.  
  3. Probably the most important advantage is the ability to run LPR algorithms on uncompressed images versus images that have been compressed. Demos of LPR technology often show a perfectly pristine license plate in a well-lit environment. What about the plate that is dirty, during a rain or snow storm, or that has a dent in it? In such cases, the image quality is key in getting an accurate read.

Here are some points I think sums it up nicely come time for a specification:

  • License plate recognition camera must be an unit with internal LPR processing  with 2 on board cameras and built in illuminator.
    • High-Resolution   Sensor XGA 1024x768 @ 30 fps, monochrome. Global shutter progressive scan CCD synchronized with illuminator flash
    • Built-in VGA color camera for video recording, overview imaging and license plate reading
    • Integrated image processor 
    • Pulsed-infrared illumination for high performance all-day license plate reading (LPR)

If you enjoyed this article and want to learn more on our AutoVu SharpX camera, check out our website. Or, simply leave us a comment and tell us what you thought about this post.


Previous Article
Some cool integrations with Security Center SDK
Some cool integrations with Security Center SDK

This week was the opening of Genetec’s Bistro in our Montreal headquarters. It’s a great new alternative fo...

Next Article
Specifying a "non-proprietary" access control system
Specifying a "non-proprietary" access control system