Hi Max,
I don't think anyone processes these so why not optimise their visualization automatically?
Yes, these maps are very useful to process integrated images. In particular, you can use them to apply a wavelet-based noise reduction. The maps have values proportional to the number of rejected source pixels. Hence, where a map has a high value that means that your integrated image is noisier, as it integrates less images.
This is probably a unique case in astronomical image processing: we know exactly where the noise is, and we know its relative strength. The standard procedure is:
- Use PixelMath to create a mask with the maximum of the dark and bright rejection maps.
- Activate the mask for the integrated image.
- With the help of STF, apply ATrousWaveletTransform for noise reduction. For example, you can simply remove the first wavelet layer as a first approximation.
This procedure will compensate for the lack of SNR due to rejection. It is particularly efficient on rejected background areas. Note that wavelet transforms are linear, so your image won't lose linearity.
Juan - perhaps we need two tick boxes next to the 'clip images' selection, to allow for inclusion of AutoSTF in the output stage.
Agreed. I'll try to implement this feature in one of the upcoming versions.
Thanks for the suggestion.
