MultiscaleLinearTransform, or MLT, is typically used for noise reduction but you can also use it for sharpening an image based on wavelet layers.

You can get an idea of what structures are at specific wavelet layers by using the ExtractWaveletLayers script, located under Script – ImageAnalysis. Select the target image you wish to use and use the default values for the rest of the settings. When you click Ok, PixInsight generates five new images – one for each wavelet layer. Inspect the images to get an idea of what structures reside at the individual layers. From this analysis, you can see that noise resides at the lower layers, usually layers one and two. The larger numbered layers contain more structure and can be targeted for sharpening using MLT.

Using MLT for Sharpening

Having inspected the wavelet layers, we’ll target layers 3-5 (with the fifth being the R, or residual, layer). Before you do that, you’ll need a mask to protect low signal areas of your image. You can create a mask in a variety of ways – for example, you could use a RangeMask, or use StarNet to create a starless version of your image, or you could use the output of the ExtractWaveletLayers script and use one of the images as a mask.

The preceding screenshot shows you the settings I used with the image of Messier 51 that we have been using throughout this series of articles. I derived the settings by experimenting with the values using the real-time preview.

The settings use the Bias setting of MLT as shown

Ensure the checkbox labeled Detail Layer is selected and for each layer set the Bias, using the images produced by the ExtractWaveletLayers script as a guide. A negative value for the Bias smooths the layer and a positive value sharpens the layer; values between 0 and +1 are typical for sharpening.

Generally, for sharpening, you can use the default settings throughout the rest of the process’s settings. The exception is the number of layers where you may wish to increase the number of layers for more control over the sharpening effect. For example, you could target only the middle layers for sharpening, leaving the other layers alone.

Conclusion

In this article, you learned about wavelet layers and learned about using the MultiscaleLinearTransform process to sharpen your images.

More Articles In This Series

This article is part of a whole series of articles about processing images using PixInsight: