Technical Application Note 0001

Guidelines for Image Acquisition for ALPR Sample Data and Applications

Revision: 1.0
Date: 2018-02-05
Contact: alpr.support@vision-components.com
Copyright: 1996-2018 Carrida Technologies GmbH, Ettlingen, Germany
Author: VC Support

Home page

Table of Contents

1   Introduction

The Carrida automatic number plate engine (ALPR) needs to be trained for each country or state to be read. For this purpose a large number of license plate image samples is needed for training and optimization of the detector.

After a sufficient number of images has been collected, the training and classification information is extracted from the sample images in a semi manual process. The images are annotated, sorted, and finally processed in order to obtain the new country specific ALPR Carrida classifier.

This document describes the properties of sample images needed for the training of a new country, including suggestions how the images can be acquired in an efficient manner.

2   Desired image sample properties

In order to train the Carrida country specific classifier system, at least 2.000 images should be provided, and of those samples no more than 4 images of a unique license plate shall be given.

The more images are provided, the better the quality of the resulting detector. Countries with license plates of different fonts require more samples to cover all font types, perspectives, and other variations.

The following image properties are required for a successful traing of the Carrida classifier:

  • All images must be sharp, without motion blur
  • No interlace in video frames, no half frames
  • The background of license plates shall not be overexposed, a slightly underexposed image is preferred over an image which is too bright.
  • In order to achieve a certain diversity in the sample image set, at least 3 different viewpoints shall be used for image acquisition of the samples
  • A mixture of street scenes and highway scenes is preferable

Hint

If using a video camera or digital photo camera, switch it to manual mode. Then set the shutter time to 3-5 ms, and manually adjust the f-stop and ISO setting to capture a properly exposed image.

3   Imaging Geometry

Since generally ALPR is done with cameras at a viewing position higher than a typical license plate, the training images should be taken with the camera placed at least 1 m above ground.

t1 t2
License plates can be viewed at a rotation angle between -20 and +20 degrees as shown on the graphic. The aspect angle of the camera relative to the car rear side or front side (sideways viewing angle) shall not exceed 30 degrees left or right, up or down.

The minimum height of characters in an image shall be at least 14 pixels for training samples (for special cases see below).

The maximum height of characters in an image shall not exceed 150 pixels for training samples.

The imaging resolution should provide at least allow a 1 pixel gap between characters in the license plate.

The minimum width of a car should be around 300 pixels. For example, for images with VGA resolution (640x480 px) the width of a car should be at least around half the image width.

./images/imageTan12.png

Maximal and minimal character height/width for traning samples.

4   License plates which contain special characters or symbols

Some countries enhance their license plates with special characters and markers to provide additional information about their use. For example China, Thailand, UAE, Saudi Arabia, the USA, and Russia enhance their plates with symbols, state signs, small strings, or stacked characters.

The resolution of images for license plate reading in those countries must consequently be higher as usual to make the shape of those special characters clearly visible. We recommend an image resolution, which results in a minimal character height of at least 25 pixels. In his case the size of the special symbols will be big enough so that they become readable.

Depending on the overall image quality, the required resolution may need to be chosen even higher to compensate for lens blur or unsharp characters resulting from motion.

./images/imageTan7.png

Good sample images (left side), and bad sample images (right side) for licene plates from UAE, Thailand and China.

On the examples above the state strings within the good license plates is clearly readable in all cases. The symbols and strings in the bad images are too blurred, or too small to be recognizable.

5   Example Images

The following images are good examples for the specifications and recommendations given in this document. They cover different imaging setups in order to demonstrate the flexibility provided by the Carrida engine.

i1 i2 i5
i3 i4 i6

Images displayed below are bad examples. Some are not sharp enough, some plates are covered by other objects or the characters are too small for recognition.

b1 b2 b5
b3 b4 b6