DACOLIAN INTRADA ALPR - Automatic Number Plate Recognition

Home > Products > INTRADA ALPR > Performance >

help

print view

INTRADA ALPR

Performance

Independent tests have identified Intrada ALPR as one of the best available license plate recognition engines. These tests, in which software of different vendors is compared using the same input images and the same processing equipment, are rare.

Performance figures based on company specific test conditions have a limited use as the actual recognition performance of ALPR depends on many different properties like the contents of the image, lighting conditions, variation in character size, camera mounting position, number of countries to be recognized simultaneously, etc.

To illustrate the quality of Intrada ALPR we will show the influence of some of these properties on the performance of Intrada ALPR. The presented figures are based on data sets containing several thousands of images. The images are captured by the application domain appropriate imaging equipment.

Application domain

Application domain and performance

Typical performance figures differ for the various application domains like parking, free flow tolling, and law enforcement. This is caused by the amount of variation present in the images. For parking applications many things are either fixed or can be measured very accurately, resulting in an almost 100% correct recognition rate.

Timing

Timing and performance

The actual processing time for recognition obviously depends on the processing power used and the image size. But also less obvious properties like complexity of the scene, minimum required recognition confidence, quality of the image will influence the processing time. With INTRADA ALPR it is possible to make an explicit trade-off between maximum processing time and recognition performance.

Character size

Character size and performance

The expected character size can be set within any Intrada application. Setting the character size to the appropriate interval will increase the recognition performance, and at the same time decrease the required processing time.