Unless instructed otherwise, Intrada ALPR will use multiple complementary algorithms to read a vehicle registration plate, so it will give the highest possible confidence level.

An internal recognition result is available after only a few milliseconds. Any additional processing time is used to improve and confirm these results before the final recognition result is generated.

For time-critical applications, the maximum available processing time can be set explicitly. This ensures that the recognition per image will never exceed the required system response time.

By restricting the available amount of processing time, recognition performance will be affected. Some typical error-correct performance curves are shown below. These curves are based on grayscale images with a size of 768x484 pixels and an average character height of 13 pixels. Some of the used images are shown here as well.

Each curve is determined using the same settings and input images. Only the maximum allowable processing time is varied. Limiting the maximum usable processing time for the recognition will not increase the error rate. As can be seen from the error-correct performance curves, only the reject rate increases.

These graphs have been made with our INTRADA Performance Analyser (IPA) tool.

The difference in performance is typical for the example images and based on a Intel P4 2GHz PC. More processing power will causes all curves to move towards the curve with no timing restrictions.

Intrada ALPR can exploit any characteristic of the input images that is fixed. Providing Intrada ALPR with more accurate estimations for characteristics like the perspective distortion, possible character height range, and region of interest (ROI), will reduce the average processing time for each image. It might even make the setting of a maximum allowable processing time unnecessary.

The current release of Intrada ALPR may yield better error-correct curves as Dacolian is constantly improving the timing characteristics of its existing recognition algorithms and implementing new more efficient techniques.