I am trying to implement the following paper: https://link.springer.com/article/10.1007/s11042-011-0835-9
In this paper, the authors say the following: 
It seems that the authors filter the image with a lowpass DWT filter and then subtract the original image with the filtered one, the outputs are supposed to be the following:
So, my questions are:
How to perform DWT 2D lowpass filtering with Python? should I generate a mask with DWT coefficients and convolve it with images? if yes, how to generate such a mask and do this operation? I checked this previous question but it does not help me. Basically the approximation of the DWT has a lower resolution than the input image and therefore I cant subtract them. So, the topic of this previous question is useless to my problem here.
I need that the filtered image must be in RGB too. How can I achieve that? filtering channel by channel or there is another smart way to do that?