AutoTRASSIR configuration comes to selection of the number of detection zones and determination of their boundaries.
Use the following guidelines during setup:
-
Number of lanes. Choose the number of lanes based on the actual width of the carriageway, indicating the nearest possible value.
Tip
The standard commonly accepted lane width of a roadway is 3,5 meters. If the camera captures the roadway width which is 8 meters (that is not only the roadway, but the entire actual width of the image in meters). In this case you should select the closest to 8 meters value, which is 2 lanes.
-
Defining identification area boundaries. Select isolated zones to obtain information about the passage of a car with a link to a particular lane (control of bus lanes, detection of car passage along the sidewalk, etc.). In addition, this will reduce the number of false alarms of the detector and will save the resources of the server, analyzing only targeted and suitable area of the image.
Important
It is necessary to take into account depth of field and number of frames shot by the camera during the passage of the car inside the zone when selecting areas of recognition. Number of frames shot by the camera will directly depend on the speed of the car. It must be remembered as well that not all frames will be suitable for recognition, the image of the license plate number should be clear and easily recognizable. In most cases, it's enough to get 4-5 frames suitable for recognition.
You can verify the settings by showing AutoTRASSIR figures. To do this, right-click on the frame and select View Options... in the context menu. Set the Show figures checkbox and select AutoTRASSIR Detector in the dropdown list.
The AutoTRASSIR figures will be displayed on the screen:
- Processing queue - This indicator reflects the state of the queue for processing license plate numbers. If the vertical bar fills up and turns red, then AutoTRASSIR begins to drop frames. The processing queue will fill if the server's CPU is heavily loaded and unable to process the frames.
- Frame recognition quality - is a rectangular indicator that displays recognition quality. Each rectangle is a separate frame used for license plate number recognition. Depending on whether or not the frame was suitable for recognition, the color of the rectangle will change from green (a "good" frame) to red (a "bad" frame).
Examples of module configuration:
-
To recognize license plate numbers of entering and leaving through the gate cars, you can select only the gate area. Cars passing on the road will be ignored in this case.
-
You should draw a separate zone on each lane of the multilane roadways.
-
In this example the camera has been installed so that the depth of field (the area of the image with the best image quality) covers only a small area in the middle of the frame. This is the same area that is relevant and useful for license plate recognition. There is no point in performing license plate recognition where the license plate is blurry or isn't visible. Limit the recognition zone to the area where the license plate is clear and of the required size.
