Video camera mounting and setup requirements for the Carmen OCR

The image received from video cameras must be sharp, contrast and readable for correct recognition of carriages and containers license plates.

Requirements for video camera characteristics:

1. Use video cameras with high resolution (720 p recommended);
2. Frame rate is not less than 24 fps;
3. The container motion speed must meet the camera operation speed. For a better recognition, it is necessary and it is enough to have 1-2 frames of a container number.
4. Possibility of setting the fix exposition value from 1/2000 s and less;
5. High sensitivity of video cameras (0,01 lx and more).

The optical axis of video camera must be perpendicular to the plane of the license plate. Use varifocal lens for settings simplicity. Select the range of focal distance changing individually for each case of system installation.

Image requirements:

1. Position of carriage or container in the frame must be horizontal;
2. Size of license plate characters in the image must be from 20 to 40 pixels, i.e. from 4% of the image height;
3. Geometric distortion by optics should be reduced.

It is also required while video camera setup:

1. Focus camera exactly to the container or carriage license plate and disable the auto focus; 
2. It is recommended to disable all the video camera forensic features. For instance, the mode of adding contrast.

The sample of mounting video camera for providing the exact recognition of carriage license plates is shown in the following figure.

In the above image, a sample application can be seen for ACCR purposes. In many cases the container codes are damaged, therefore we suggest to capture multiple images from different sides of the container and handle them together as an image sequence. Three container cameras are recommended to reach the most precise reading results. To provide optimal conditions for taking an image, use two external white LED illuminator devices. With the help of an additional ANPR camera, the truck, carrying the container, can be identified.

To reduce consumption of the computer resources use external triggering to capture images only if the container code and the license plate are in a good position.
The inductive loop (other triggering possibilities: laser sensor, microwave sensor, magnetic sensor or infrared barrier) serves as a sensor, which indicates the arrival of the vehicle.
As the vehicle stops at the “STOP” sign, all cameras identify the container code and the license plate of the truck.

The reference image received from correctly mounted and setup video camera is shown in the following figure.