License Plate Detection Technique

Following it is introduced the self-triggered plate detection technique implemented at Imagsa’s cameras:

1. Virtual Loop:

Capture 250 images per second on a fixed virtual loop.

2. Multi-exposition:

Every image is captured with a different exposition time to overcome global and local illumination challenges on the license plate region.

3. Image analysis:

An FPGA-based image processor analyses hundreds of images per second to detect a license plate pattern.

4. Segmentation:

The region of interest is segmented from the overall picture, so to speed up the OCR engine.

5. Multi Streaming:

Along with the segmented license plate image, the camera provides a scene image, optional video and other data of the detection (time stamp, GPS location, camera name, ...).

Benefits:

• Space Resolution:

Detect the license plate at the same point in the road, always.

• Pixels Resolution:

Exact 25 pixels/character, to achieve top recognition accuracy.

• Best illumination:

Through an innovative adaptive multi-exposition technique.

• No False detections:

Plate detection technique it is not affected by shadows.

• Real-time results:

Every plate is detected & segmented in real-time.

• Multi Streaming:

Plate image, Scene image and Video.