Skip to main content

Improve LPR Read Rates

WhatThe Improve LPR Read Rates guide provides administrators and enforcement personnel with instructions on how to physically and digitally optimize License Plate Recognition (LPR) cameras. Its primary purpose is to help increase the opticalaccuracy rangeand speed of plate reads through physical adjustments, exposure tuning, bounding box modifications, and system configurations. This article is intended for theOPS-COM reader?administrators and enforcement staff.

Setup and Configuration

Optimizing your LPR read rates involves a combination of physical camera adjustments, web interface configuration, and handheld settings.

Physical Camera Setup

  • Optical Range: The sweet spot for plate reads is between 1.5 to 2.5 meters between the camera and the license plate.

    Also
  • check
to

VJUimage.png

make
    sure
  • Camera Angle: Ensure the lens is clean. Another issue may be camera angle.  The optimumoptimum angle for the camera is straight on (parallel) to the plateplate.

    image.png

  • Fuzzy Searching

  • The PL8-RDR system supports fuzzy searching based on user supplied values.

    In the case of the LPR system "seeing" ABC123, it will also validate against A8C123Maintenance: andAlways ABC128check orto A8C12B.

    make

    Manual Corrections

    You can improve plate reads by making corrections and verifying correct plates on the system. OPSCOM has an algorithm that includes the human updates as part of the read score.  Therefore, if a plate is misread and corrected, it is very probable that the system will return the (human) updated version.

    Further, the system does fuzzy searching... this means it will look for a plate in the system that matches a pattern.  For example, a plate of ABC123 may be read as A8C128.  With fuzzy searching and based on time of day etc. the system will return ABC123 if that plate has a permit (but will also note that it was a "fuzzy" read of the plate).

    If you see a score of 100% in the LPR report that means that the system determined a human updated version was the best match since a score of 100% is theoretically impossible with LPR cameras.

    As an Example here's a tweak you can do for Ontario users oncesure the camera lens is setup..completely clean and free of debris.

Web Interface Configuration

Administrators can use the web interface to fine-tune how the camera processes images and characters.

Adjusting Character Dimensions: You can tweak character dimensions to improve regional reads (for example, Ontario plates).

  1. GoNavigate intoto the Web Interface Page and select Plate Reader then Camera OCR and set the following Characters dimensions:.

  2. Set CHAR SIZE Min W:W 10to Max W: 50 Min H:30 Max H: 10010.

    image.png

  3. Improving Exposure

    For dark plate reads:

  4. Set CHAR SIZE Max W to 50.

  5. Set CHAR SIZE Min H to 30.

  6. Set CHAR SIZE Max H to 100.

AsGimage.png

Improving Exposure for Dark Plates:

  1. Set Max Shutter to 500
    500.

  2. Set Max Gain to 256256. (If images are too darkdark, this setting can be increased.increased, Howeverbut it is not recommended to go overexceed 1000.)

  3. Set Iris Level to 15

    15.

image.png4Pdimage.png

Adjusting the Bounding Box

Box:

The LPR camera has the ability to customize the area that is being analyzed during the plate read. This article explains how to adjust the read area (Bounding Box).
The bounding box refers torestricts the area that can be adjusted to narrow the view orspecific area that is analyzed by the LPR camera.camera,

ignoring


Theunnecessary settingsbackground involved are located in the Plate Reader Settings.

image.png


The default settings for full frame are as follows.information. Measurements for height are always considered the distancecalculated from the top of the screen.screen, Measurements forand width are based on distance from the left of the screenscreen.

9ISimage.png

  1. Navigate to the Plate Reader Settings.

  2. Edit the pixel values to narrow the read area. The default full-frame settings are:

    • Win MinX Pixel -: 0

    • (From
    • the left)

      Win MinY Pixel -: 0

    • (from
    • the top)

      Win MaxX Pixel -: 1279

    • (From
    • the left)

      Win MaxY Pixel -: 1023

      (from
  3. Increase the top)

    Min

    image.png

    values


    Toor viewdecrease the locationMax ofvalues to shrink the bounding box gohorizontally into(X) or vertically (Y).

  4. Click the CameraApply OCRbutton tab.

    to

    image.png



    Here issave the full frame view of the bounding box. See the red outlined box in the image below.
    changes.

    image.png

  5. Adjusting Height

To adjust the box edit the number in the settings. In this case we will narrowOnce the bounding box in height to restrict the read area.

Here are the settings:

Win MinX Pixel - 0 (from the left)
Win MinY Pixel - 100 (from the top)
Win MaxX Pixel - 1279 (from the left)
Win MaxY Pixel - 700 (from the top)

image.png

IMPORTANT

Once the settings are changed and you click onthe Apply,Apply button, the camera will reboot. This process will take a couple of minutes to complete before you can navigate away from the page.


Handheld Unit Configuration

TheFor resultingNOREAD Bounding Box will look like this:

image.png

Let's now say we want to lower the box in the view. I would have to increase the values in both Y axis settings.

Here are the settings

Win MinX Pixel - 0 (from the left)
Win MinY Pixel - 200 (From the top)
Win MaxX Pixel - 1279 (from the left)
Win MaxY Pixel - 800 (From the top)

image.png

Here is the resulting bounding box:

image.png

Adjusting Width

Additionallyresults, you can narrowadjust settings directly on the widthAndroid ofhandheld thedevice boundingto boxprocess ifvanity needplates be.or Inchallenging this case we will narrow the bounding box by 100 pixels on both sides.reads.

Here

    are the settings

  1. Win MinX Pixel - 100 (from the left)
    Win MinY Pixel - 200 (From the top)
    Win MaxX Pixel - 1179 (from the left)
    Win MaxY Pixel - 800 (From the top)

    image.png



    Here is the resulting bounding box.

    image.png

    By selective use of these settings you can customize the area that is processed by the LPR read. As a result LPR reads could be faster and more accurate since the camera is reading a smaller more focused area and ignores anything outside of the bounding box.


    On the Handheld

    For NOREADs you can check the setting inTap System Settings.

    to
  2. see
  3. if that can be reduced.  
    It may mean more noise, but it may also help with determining what is going on (ie. reading but not seeing it as a

    Tap READ result)

    To adjust this setting go into system settings and tap on Include NOREAD (ie. vanity) Results

    .

    image.png

  4. Set the percentage to 50% to start.

    We
  5. can
  6. tweak

    Tweak itthis oncepercentage welater seebased on how the system reacts.reacts in the field.

Using this Feature

The LPR system utilizes advanced algorithms and human corrections to continually improve its read accuracy during daily operations.

Fuzzy Searching

The PL8-RDR system inherently supports fuzzy searching based on user-supplied values. If the LPR system "sees" ABC123, it will automatically validate against similar visual patterns such as A8C123, ABC128, or A8C12B based on time of day and active permits.

Manual Corrections

You can actively improve plate reads by making manual corrections and verifying correct plates on the system. The system uses an algorithm that includes your human updates as part of its read score. Therefore, if a plate is consistently misread and corrected by an administrator or guard, it is highly probable that the system will return the human-updated version in the future.

If you see a score of 100% in the LPR report, that indicates the system determined a human-updated version was the best match. A raw algorithmic read score of 100% is theoretically impossible with standard LPR cameras.


Best Practices and Considerations

  • image.pngOptimize the bounding box: By selectively adjusting the bounding box settings, you can heavily customize the area processed by the LPR read. Focusing the camera on a smaller area forces it to ignore surrounding noise, resulting in faster and significantly more accurate reads.

  • Balance NOREAD noise: While increasing the NOREAD percentage on the handheld can help identify vanity plates, it may also introduce more system noise. Monitor the results and find a balance that captures valid plates without overwhelming the user with false reads.

  • Avoid excessive gain: When adjusting exposure, never increase the Max Gain setting over 1000. Excessive gain will introduce heavy digital noise, actively degrading the camera's ability to perform OCR (Optical Character Recognition).