Carrida Camera Installation Guide

Carrida Camera Installation Note for access control

Revision: 1.2
Date: 2017-08-30
Contact: alpr.support@vision-components.com
Copyright: 1996-2017 Vision Components GmbH Ettlingen, Germany
Author: VC Support

Table of Contents

1   Introduction

This manual describes the installation of the Carrida Camera for access control and parking.

See also the related documents for further details and technical information:

A list of abbreviations used throughout this document:

ARM
Stands for processors with the ARM instruction set.
ALPR
Automatic License Plate Reader.
dB
Decibel, in the Carrida documents used to describe a relative sensitivity (gain) value in cameras.
FOV
Field of view.
HW
Hardware.
LP
License Plate.
ms
Millisecond, one 1/1000 th of a second.
PC
Personal Computer.
px
Pixel. The smallest unit of resolution in an image. For example, VGA images are made up of 640x480 pixels.
ROI
Region of Interest, typically a marked area in an image.
SDK
Software Development Kit.
SW
Software.
x86
Stands for Processor Architectures with the Intel instruction set.
VC
Vision Components GmbH.
OSD
On-screen display
Image symbols used in this document
Symbol Meaning
Note Sign Points to especially important aspects of the manual or the software.
Hint Sign The Light bulb highlights hints and ideas that may be helpful.
Warning Sign This warning sign alerts of possible pitfalls to avoid. Please pay careful attention to sections marked with this sign.

2   Carrida Camera properties

The Carrida ALPR Camera is equipped with a 12 mm lens, which is suited for most access control applications. The viewing angle with this lens is 30° horizontally and 20° vertically.

The working distance of the Carrida Camera ranges from 2-8 m. Table 1 below shows the field of view (width of observable area) depending on the distance of the vehicle from the camera.

Table 1: The field of view at different distances.
Distance Field of View  
2m 1,10m wi
   
4m 1.55m
   
5m 2.07m
   
6m 2.59m
   
7m 3.11m
   
8m 4.14m

For distances above 4 m, we recommend to switch the camera to high resolution mode. This doubles the resolution of the camera and accordingly enhances the reading capability. You can change this setting in the Camera Configuration page by turning off the low resolution flag (see |car| Camera Program User Documentation).

3   Getting Started

Install the camera according to the given limitations:
  • The viewing angle from the camera to the license plate should not exceed 30° in any direction.
  • The field of view of the camera is 30° horizontally, and 20° vertically (see also table above).
  • The camera should be mounted at least 2 meters from the expected closest position of a license plate.
  • Plates can be rotated with an angle as much as 20° clockwise or counterclockwise.

For best reading results, we recommend to mount the camera on a minimum height of ca. 80cm. This avoids the disturbance of the vehicle headlights shining directly into the camera, which otherwise might reduce the image quality when the vehicle is very close to the camera.

After mounting the camera, connect the power and the LAN cable. Open a web browser (e.g. Opera, Firefox) and use the default IP address in order to access the configuration page of the device.

Note

Note Sign The default IP-address of the camera is 192.168.3.15.

On the configuration pages of the camera you can modify its setup and view a live image and reading results.

During and after configuration, the best practice is to test the ALPR: drive a vehicle into the position where it should be possible to read the plate and check the reading results.

3.1   Connecting the Camera

The camera is supplied with 2 cables with a length of 5m. One cable is used for connecting the camera to the network (green), and the second cable is needed for power supply (black).

./images/pro_z_abmessungen.png

Figure 1: Pro Z Camera

For power supply and I/O signals the Carrida Camera uses the J0 connector. Its pin assignment is shown in he following Table 2.

Table 2: Pin assignment of J0connector  
Camera socket rear preview Pin Signal Level Cable standard color
w 1 Main power supply +12-24V Brown
2 Common ground GND Blue
3 PLCIn0 +12V White
4 PLCOut0 +12V Green
5 PLCIn1 +12V Pink
6 PLCOut1 +12V Yellow
7 PLCIn2 +12V Black
8 PLCOut2 +12V Gray
9 PLCIn3 +12V Red
10 PLCOut3 +12V Purple
11 FLASH0/TrigOut +12V Grey/Pink
12 FLASH1 +12V Red/Blue

Warning

Warning Sign Make sure that the current is always appropriate to the limitations. If exceeded, the FLASH0 and FLASH1 lines could be destroyed.

More detailed information about hardware specifications can be found in the VC pro Z Series Operating Manual .

4   Mounting the camera

The camera should be mounted and stabilized on a place which is least 2 meters away from the closest position where the vehicle is expected. Closer distances can work, but the reading accuracy may be less in those situations.

The best recognition distance is between 2 and 8 meters. The camera should be installed by keeping in mind the angle and distance limitations as described in following paragraphs.

4.1   Limitations to be considered

The camera has a field of view (FOV) of 30° (A) horizontally and 20° (B) vertically, as shown in Figure 2.

./images/CameraInstallPic17.png

Figure 2: The field of view of the Carrida camera when equipped with the standard 12 mm lens.

As further consideration, the angle between the plate surface and the camera should not exceed 30° in any direction in order to obtain the best reading results. Figure 3 gives an example of an installation where the camera reaches the limit of 30° horizontal (A) and 30° vertical (B) viewing angle.

./images/CameraInstallPic14.png

Figure 3: Requirements for the camera position.

Lastly, the rotation angle of the license plates must not be greater then 20° clockwise or counterclockwise as shown in Figure 4.

./images/CameraInstallPic02.png

Figure 4: The angle of the plates must not exceed 20° clockwise or counterclockwise.

4.2   Mounting directly at the gate

The Carrida camera can be mounted directly by the barrier pole, either in front or behind it, with an unobstructed view of the vehicle plates. If the camera is mounted directly at the gate, it can be positioned at bumper level height (ca 0.50 m). The horizontal angle in this case will be reduced to the minimum, as shown Figure 5.

Make sure that the camera is protected from the vehicles, and has an unobstructed view, without any objects that could get in the way.

./images/CameraInstallPic9.png

Figure 5: ALPR mounted on a bumper level behind the gate (A), and in front of the gate (B).

We recommend for the Carrida camera to be mounted a few centimeters above the level of head lights of incoming vehicles. This reduces the glare of the headlights into the lens and thereby minimizes flares, which could affect the image quality. In the following Figure 6 the camera is positioned with a slight angle to the plates (15°), above the direct beam of the headlights.

./images/CameraInstallPic11.png

Figure 6: The Carrida camera mounted on a pole above the vehicle headlights at ca 80cm height above street level.

4.3   Mounting on a pole / elevated position

It is possible to mount the Carrida camera on a higher position, as shown in Figure 7, but this may limit the useful range within which license plates can be read.

./images/CameraInstallPic5.png

Figure 7: The Carrida Camera mounted on a 2.2 meter high pole behind a barrier.

From this perspective and height, the Carrida camera recognizes vehicles resp. their plates, which are more then 4 meters away from its position. Tilting the camera further down would make the viewing angle to the plates too steep (more then 30°), and as a consequence decrease the reading performance. See the following Figure 8.

./images/CameraInstallPic8.png

Figure 8: Vehicle approaching closer to the barrier.

5   Setting up the camera

After connecting the camera to a PC or laptop, open a web browser and enter its default IP-address 192.168.3.15 in the address bar (the IP address of the camera can be changed in the setup).

The main page of |car| Camera Configuration will open as shown below.

./images/CameraSoftwarePic1.png

Figure 9: The main Carrida camera configuration page

The |car| Camera Configuration page has the following menus:
  • Camera configuration
    • display formatting
    • gain and shutter settings
    • defining the ALPR-autolum region
    • brightness settings
  • IR flash configuration
    • enabling/disabling the flash
    • flash power settings
  • ALPR Configuration - trigger, stream and engine settings

  • access Configuration (access control, list location)

  • event Configuration (read and no-read events)

  • system configuration (user information, network settings)

  • ALPR results

  • text results

  • device information

Detailed information about how to setup the Carrida Camera using the configuration pages can be found in the |car| Camera Program User Documentation.

5.1   Adjusting the camera exposure control

Every camera setup is different and the automatic brightness control of the camera needs to be adjusted according to its mounting position and viewing angle with respect to the lane or barrier. The autoluminance parameters are used for this purpose. It is important to properly set these values, as they will substantially influence the quality of the license plate reading results by avoiding overexposed or underexposed license plates.

The camera automatically controls its exposure based on a brightness measurement in the live video, constrained by minimum/maximum values for exposure. The most important setting is the area, in which the camera measures the current image exposure - this area should be only the region of the image, which can contain a license plate. The image area, which is used for measurement and control, is called the autoluminance-polygon.

In the image display, the autoluminance-polygon is drawn by green lines. The polygon corners can be moved by clicking and dragging them to the desired location in the image.

It is important that this region has the smallest possible dimensions, so that it only covers those parts of the image, where actually license plates will pass through.

Select the area where you expect the vehicle plates, like shown in the image below.

./images/CameraInstallPic19.png

Figure 10: Limit the autoluminance-polygon to the image area, which can contain license plates.

Under the Camera Configuration parameter set you can find settings for the Camera and the LED flash. The autolum parameters are located in the Camera Settings tab.

./images/CameraInstallPic18.png

Figure 11: The autoluminance parameters used for exposure control.

We recommend to enable 'Use Anpr-autolum'. By checking this box you will activate the camera exposure control within the autoluminance-polygon.

The exposure control of the camera counts all good pixels within the autoluminance area. Good pixels are those, which are not too dark or to overexposed.

In the cases where there are too many overexposed pixels or to few good pixels, and the camera will change its exposure settings to make the image brighter or darker. The correct pixel counts are indicated as percentage (0-100) relative to the image size.

Hint

Hint Sign What to do if the license plates are always overexposed?

If the license plates are constantly to bright, the percentage value of maximum saturated pixels needs to be reduced, so that the camera tries to reduce the count of those pixels. Example: reduce the value from 5 to 3 % and observe the effect in the live image.

Hint

Hint Sign What to do if the license plates are always too dark?

If the license plates are constantly to dark, the value of minimum correct pixels needs to be increased, so that the camera tries to increase the number of those pixels. Example: modify the value from 3 to 5 % and observe the effect in the live image.

5.2   Basic Event configuration

The Carrida Camera can be set up for specific actions, which take place after certain events (reading, IO signal)occur. For example when a license plate has been read, the camera can automatically send the current license plate images to a specified ftp server.

To define actions for events, go to Event Configuration page. Carrida supports three types of events, which can trigger actions:

  • read event (a LP has been read)
  • no-read event (no LP has been found within a defined time limit)
  • low certainty event (a LP has been read, but with low confidence)

Details to the above mentioned event types can be found in the |car| Camera Program User Documentation*l. For typical access control applications, only *read events are used.

Actions can be defined on the event drop down lists in configuration page - click on ADD, and a list of actions will appear. Select any action and it will be added, as shown in Figure 12 below:

./images/CameraInstallPic21.png

Figure 12: Setting up a Ftp server.

Seven different actions are available, the most common being ftp Upload, and Ftp database. The former action executes the transfer of the last read image to a remote ftp server, the latter updates a csv file with all readings on a server.

  • Ftp Upload
  • Ftp Database
  • Store
  • Digtal Out
  • Notification
  • Serial
  • TCP
Setting up the ftp server for image transmission is done as follows:
  • Enable the action.
  • Enter the IP address of the host.
  • Enter the credentials in the following form: [username]:[password].
  • Assign a naming pattern which will be used for the filenames saved images.
  • Select 0. no image uploads, 1. for full image upload, 2. for license plate image upload only.
  • Define the image quality for sent images (recommended 70%).

The option Ftp Database has a similar set up to the ftp upload action. When triggered, the Carrida Cam will create and update a csv file on a specified host with information about readings.

To create an Ftp database action:
  • Set enable to true.
  • Enter the IP address of the host.
  • Enter the credentials in form of: [username]:[password].
  • Define a file where readings will be saved to.
  • Assign a naming pattern under which the readings will be saved.

Actions can appear multiple times, and they can be combined with other actions. A detailed description of possible actions and their parameters can be found in the |car| Camera Program User Documentation, Chapter 7.

Naming patterns enable user defined filenames, which contain the date, time, plate numbers, and other information about a specific reading. The most common placeholders are:

%LP% - license plate %CONF% - plate confidence
%STATE% - state %CONF_STATE% - state confidence
%DAY%.%MONTH%.%YEAR% - date %HOUR%:%MINUTE%:%SECOND% - time

Characters within the percentage signs (%) are placeholders for data that is specific to each reading, and will be replaced when the result is generated.

Here are some examples of how formatting with pattern works, for a German license plate B1234CV, which has been read on July 25th, 2017 at 7:45:25 with 100% plate certainty and 95% state recognition certainty.

%LP%,%CONF%,%STATE%,%CONF%,%HOUR%-%MINUTE%-%SECOND% B1234CV,100,DE,95,07-45-25
%LP% from %STATE% detected on %DAY%.%MONTH%.%YEAR%. B1234CV from Germany detected on 25.07.2017
%LP% was detected at %HOUR%-%MINUTE%-%SECOND% B1234CV was detected at 07-45-25

Placeholders will be replaced with real data, and plain text will remain the same. File names can be formatted in many different ways using patterns whose complete list can be found in |car| Camera Program User Documentation, Chapter 12.

6   Parking Mode

The Carrida Camera can be configured so that it either transmits every reading result directly to a client or to a connected application. In many situations this behavior is undesirable, because only one good result is sufficient.

In parking mode, the camera is set up so that only one filtered and verified reading is transmitted if a new license plate enters the field of view.

Note

Note Sign The parking mode is enabled by setting the parameter Drop After to a value > 0. If it is set to 0, every reading will be transmitted by the camera.

How the parking mode works:

The purpose of the parking mode setting is to create a filtered unified reading for multiple readings of the same license plate. Only this reading is then transmitted to a server.

When the Carrida Camera is operated in parking mode, it scans its field of view and tries to find license plates until a certain defined number of valid readings has been made. It then transmits the reading and waits until this license plate disappears, and at least Drop After readings with a different or no license plate have occurred.

The parameter Minimum Detections controls how many readings of the same license plate have to be made until it is accepted. Only after this number of readings the transmission of a result will take place. A new reading is accepted as valid if

In parking mode it is desirable to transmit a reading only once, but the camera must be ready to accept new license plates after a while. The parameter Drop After defines the number of readings, respectively the time span after which a new reading can be transmitted again. The Carrida Camera has a read cycle time of approximately 250ms. A value of 4 for Drop After therefore corresponds to about 1 second.

The Buffer Size sets the internal image buffer, which is kept in camera memory before transmission. The purpose of this buffer is to store images and potentially correct readings before they are transmitted. The transmission of images and reading results starts after at least Buffer Size images have been captured. A larger Buffer Size may lead to a delayed transmission (Buffer Size * Camera Cycle Time, see above), but it has the benefit that multiple readings can be merged into a better output result.

Note

Note Sign A buffer size > 1 delays the output of images and readings. This can sometimes not be desired for some access control scenarios.

The same holds also for the Minimum Detections value, it will also delay the transmission at least by the value set for it.

ALPR parking mode examples

The following examples clarify the use of the parking mode settings in more detail. Assuming that two vehicle enter the field of view of the camera, the following events may take place.

Example 1: For a setting of Minimum Detections = 2, Buffer Size = 1, Drop After = 2, Maximum Plate Diff = 1

Time Reading Confidence Camera Action
09:00:02.0 123 40  
09:00:02.3 AB123 50  
09:00:02.6 ABC123 60 the camera has a valid result and transmits ABC123 with a confidence of 60
09:00:02.9 ABC123 50 all further readings will be ignored, no transmission
09:00:03.2     no license plate found
09:00:03.2     no license plate found
      no license plate found
09:00:08.1 OS456 50  
09:00:08.4 OS456 70 the camera has a valid result and transmits OS456 with a confidence of 70
09:00:08.7 OS456 60 all further readings will be ignored, no transmission

Example 2: For a setting of Minimum Detections = 3, Buffer Size = 1, Drop After = 1, Maximum Plate Diff = 1

Time Reading Confidence Camera Action
09:00:02.0 123 40  
09:00:02.3 AB123 50 difference is greater than 1 to previous reading, first new reading
09:00:02.6 ABC123 60 difference = 1, second reading
09:00:02.9 ABC123 50 difference = 0, second reading, the camera has a valid result and transmits OS456 with a confidence of 60 (the biggest confidence found)
09:00:03.2     no license plate found
09:00:03.2     no license plate found
      no license plate found
09:00:08.1 OS456 50  
09:00:08.4 OS456 70 the camera has a valid result and transmits OS456 with a confidence of 70
09:00:08.7 OS456 60 all further readings will be ignored

A Buffer Size > 1 basically only delays readings by the number of images captured (with the cycle time of approximately 300 ms) and smooths ouyt reading errors, which have been compensated (like missing characters). A buffer size > 1 is therefore mosty useful for video processing and visualization. For access control it should be set to 1. The working of this parameter is demonstrated in the next example below:

Example 3: For a setting of Minimum Detections = 2, Buffer Size = 3, Drop After = 2, Maximum Plate Diff = 1

Time Reading Confidence Camera Action
09:00:02.0 123 40  
09:00:02.3 AB123 50  
09:00:02.6 ABC123 60 the camera has a valid result and transmits ABC123 with a confidence of 0.6
09:00:02.9 ABC123 50 all further readings will be ignored, no transmission
09:00:03.2     no license plate found
09:00:03.5     no license plate found
      no license plate found
09:00:08.1 OS456 50  
09:00:08.4 OS456 70 the camera has a valid result, but does not transmit because the buffer size (3) is not yet reached
09:00:08.7 OS456 65 image OS456 with confidence 70 is transmitted
09:00:09.0 OS456 64 all further readings will be ignored, no transmission