- Brightness control
When sudden changes in brightness occur in the scene, the difference between the current frame and the background suddendly becomes extremely high, generating noise on the foreground mask. This anomaly stops for a few seconds the processing, allowing the background to automatically adapt to the change in brightness of the scene.
Due to efficiency reasons, the algorithm operates on a grid built on the image and examines the differences in brightness only in the sections of the grid.
You can set the following parameters:
- Grid Thickness: thickness of the grid used to detect the change in brightness. A low value increases algorithm efficiency with a small loss in terms of accuracy. Setting an high value generates a dual situation.
- Sensitivity: by increasing the value, the algorithm becomes more sensitive in detecting changes in brightness. The suggested value is 0.1.
- Threshold: Threshold for detection of changes in brightness. By reducing the value, the algorithm becomes more sensitive to changes in brightness. By setting the value at 255 you deactivate the mechanism.
- Enable control of "night mode": if enabled, detects night/day and day/night transitions and shortly stops the processing to adapt to the change in brightness.
By setting these parameters the algorithms become more efficient.
- Spatial decimation: consists in diminishing the resolution at which the algorithm processes the image. You can reduce the dimension by a factor of 2 or 4, elaborating an image that is respectively one quarter or one sixteenth of the original one.
- Time decimation: allows to discard a certain amount of frames, processing an image each X milliseconds.
- ROI: draw a Region of Interest (smaller than the framed image) where to perform the elaboration.
N.B. These operations are crucial when you can use only the primary stream of a camera and both the resolution and the frame rate are high, thus not allowing the live processing from the application.
- Resolution and Rotation of the image
In the drop-down menu you can set the following parameters:
- Device name: desired name of the device
- Resolution: you can leave the default setting or customize it as well. When selecting the resolution it is crucial to make sure that the camera can support it.
- Rotation (degrees): Not every camera can acquire a rotated image. This operation is particularly useful when, for instance, you can install a camera in portrait mode, so to use the horizontal angle of view of the camera to frame a standing person. For this very reason A.I. Tech products allow the processing of a 90°, 180° and 270° rotated image.
N.B.: This parameter can only be configured on EDGE plugins.
- Shadow removal
The "Shadow removal" analyzes differences between background and frame colour.
The algorithm is based on the analysis of the difference in chromaticity between the background and the current frame, as the shadows generally make pixels darker.
Reducing the "Minimum Brightness Ratio fg/bg" or increasing values "Maximum colour difference" and "Maximum increase in saturation", the algorithm becomes more sensitive and removes darker shadows as well. Indeed, it could happen that no shadow pixel at all is removed.