Unwanted camera features

When you configure cameras for LPR, note the following:

Automatic gain adjustment

One of the most common types of image interference caused by cameras is gain noise.

  • Gain is the way that a camera captures a picture of a scene and distributes light into it. If light is not distributed optimally in the image, the result is gain noise.

    Example of image affected by gain noice.

    Controlling gain requires that complex algorithms are applied, and many cameras have features for automatically adjusting gain. Unfortunately, such features are rarely helpful in connection with LPR. Milestone recommends that you configure your cameras’ auto-gain functionality to be as low as possible. Alternatively, disable the cameras’ auto-gain functionality.

    In dark surroundings, you can avoid gain noise by installing sufficient external lighting.

Automatic enhancement

Some cameras use contour, edge or contrast enhancement algorithms to make images look better to the human eye. Such algorithms can interfere with the algorithms used in the LPR process. Milestone recommends that you disable the cameras’ contour, edge and contrast enhancement algorithms whenever possible.

Automatic compression

High compression rates can have a negative influence on the quality of license plate images. When a high compression rate is used, more resolution (see Plate width recommendations) is required in order to achieve optimal LPR performance. If a low JPEG compression is used, the negative impact on LPR is very low, as long as the images are saved with a JPEG quality level of 80% or above, and images have normal resolution, contrast and focus as well as a low noise level.

Acceptable

Unacceptable

Example of the use of automatic performance with acceptable LPR quality. Example of the use of automatic performance with unacceptable LPR quality.

License plate image saved with a JPEG quality level of 80% (i.e. low compression).

License plate image saved with a JPEG quality level of 50% (i.e. high compression).