Detection Masks
The default detection mechanism in indi-allsky selects a rectangular region in an image to perform star and meteor detection. This works well when you have an unobstructed view, but this is not always the case.
The detection area for stars and meteors should be unobstructed. Gaps in tree leaves can cause false positives in the star detection. Contours of trees and houses can be detected as lines in the meteor detection.
Masks allow you to more specifically crop areas of an image.
indi-allsky generates a new base mask image every time the program is started, once a stable exposure is found. You can download the base image from the Tools -> Mask Base view.
You may notice the Mask Base image is flipped from the normal images generated by the system. This is normal and expected. The detection mask needs to be based on the original image that received by the camera, before the image is flipped, scaled, rotated, etc.
Using the following image as an example.
The default Region of Interest [RoI] for this image looks like this. The central region is automatically calculated based on the dimensions of the image.
Using a Mask, arbitrary boundaries can be set to exclude/include only specific areas of the image. White areas of mask are retained, while black areas are masked out.
Mask:
Using this mask, the resulting detection area looks like this.
Note: The edges in the mask need some distance between contours of obstructions and the edge of the mask due to the blurring that is applied for line detection. This mask is exaggerated just for demonstration purposes.
A straight line in the mask will be detected as an edge/line in the meteor detection code. indi-allsky will blur the edges of the detection mask to prevent false positives due to the mask.
