How to add wavelets after extraction of wavelets? (Jedi question)

[madrid sky]

Hi folks,

This is a question for you jedis of the force!! (The force may be with you). Let's say I want to extract wavelets, deal with them individually and then recover back original image. I know how to "extract wavelets" through the Script under "Image Analysis" called "Extractwaveletlayers".

But now... is there any script or way to add those wavelets back and recover original image?

I have tried "Imageintegration" process but it doesn't seem to go well. I don't know how to make it work.

Thank you in advance!
Sam

[msmythers]

Sam

Here is a simple method I think. Using MultiscaleLinearTransform and PixelMath. I used MLT to extract a selection of wavelets. I then used PixelMath to put them together. To get the exact same image as the original I would many more wavelet layers but you can see how well this works. Of course I could do whatever I want to the individual wavelets if desired and blend into the original image or use the wavelet as a mask. The other thing is you can also work on the individual wavelet in MLT. It's and extremely powerful tool.


Mike

[madrid sky]

Thank you Mike! So kind of you. But I am not used to Pixelmath and it doesn't seem to work good. It asks to apply it on an image, and when I do it, result is not good. I just want it to create image from scratch given the images stated under "expressions". Once I tell Pixelmath images to join, what should I exactly do? How do you extract wavelets from MLT? Why not using "extractwaveletlayers?

The idea is to do deconvolution on individual layers and then join them together. In fact, I don't want to use MLT. I want to test deconvolution in depth. In short. I want to deal individually which each layer, apply deconvolutions to each and then put them back together. Then from that point on, I will use other processes... thing is I could extract layers, but not put them back together!! I am sure Pixinsight has some option for it.

I wait for an answer.

[madrid sky]

Sam

Here is a simple method I think. Using MultiscaleLinearTransform and PixelMath. I used MLT to extract a selection of wavelets. I then used PixelMath to put them together. To get the exact same image as the original I would many more wavelet layers but you can see how well this works. Of course I could do whatever I want to the individual wavelets if desired and blend into the original image or use the wavelet as a mask. The other thing is you can also work on the individual wavelet in MLT. It's and extremely powerful tool.


Mike

Wait a minute... I think I understand now how you did it. Let me test it again... I will be back.

[madrid sky]

To Mike:

I don't get it working. In fact, I am using 4 layers (using MLT, I create five, including the "base layer"), but when I combine them back (I join 4 layers through Pixelmath and I apply them to "base layer"), result is not the same as original. I am missing something. Let me know how you do it.

See attached file.

[msmythers]

Sam

Look at these links. I think they might help your understanding of wavelets.

https://pixinsight.com/doc/legacy/LE/20_wavelets/understanding_multiscale/understanding_the_multiscale_language.html

https://pixinsight.com/doc/legacy/LE/20_wavelets/wavelets_processing.html

They go back a long way to Pixinsight LE in 2005.


Mike

[madrid sky]

I have been using wavelets for decades. My problem is other. I want to do individual layered deconvolution (not MLT processing), and Pixelmath doesn't seem to be working good.

1) When I add back those 5 pictures captured with MLT by "snapshooting" the preview (maybe this is my mistake)
2) and try to get them together back with Pixelmath,
3) Result is not as the original.

What is going on?