The OCR engine of our software will detect ANY pre-learned character (based on License plate design) and extract them from any captured image through powerful and optimum algorithms.
This system is not limited to different types of plate. For example, any change in character type (different country), plate color (with, yellow, red, etc.), forward or backward plate mounting and even changes in location and size do not have an improper effect on the result of the engine.
According to the real application output which was reviewed before, accuracy of the system ‘s OCR engine is roughly 97%, therefore this system obviously achieves the top position among other competitors. Also, the License plate reading is independent of the vehicle’s speed and till now it is used for vehicles with speeds up to 200 km/h.
Iranian license features
We can point that according to the application of this engine for Persian license plates, it achieved some unique abilities. For example, some characters in Iranian plates are very similar such that sometimes it is even hardly recognizable for humans, but this system distinguished them from each other with such high accuracy. Some of these conditions are referred here:
Light and weather condition
Based on powerful hardware and software design and the application of powerful internal and external illuminators, the system is not dependent on the light condition of the environment, so it works conveniently all day long and in any weather condition.
Pictures above demonstrate the independency of system to the light condition
Achieving 97% accuracy in real condition is not easy and cheap!
ANPR in tough situation
The ability of detection and reading some defective plates by ANPR core is very unique and fundamental, and this feature is demonstrated in the pictures below:
Dirty license plate
License behind a net
License behind the front window
Multi license plate vehicle
Partial overlapping of vehicles