Rolling Guidance Filter

edit

Yeah I went through the paper, as you will see they have used Joint Bilateral Filter in the process. And in the source code present over (http://www.cse.cuhk.edu.hk/leojia/pro...) , they have tried filtering using Permutohedral lattice. So is there any way any of those filters used can be found in opencv which takes two images as base for input.

@berak Do you know of any such filters? I have posted as a new question : http://answers.opencv.org/question/16...

unfortunately i don't (but i'm looking into it).

looking at your other questions, i'm just thinking, that if at all, this should be build in c++ (and added to ximgproc), so python or java folks can use it, too.

did you take a look at their implementation ?

@berak Yeah I looked into the source code. They have not used JointBilateral there but instead used bilateral filter by using permutohedral lattice. In XimgProc I see JointBilateral but how to use in Java in an already created project with opencv as its just there for Python. http://docs.opencv.org/3.0-beta/modul...

btw, their permutohedral code seems to be from http://graphics.stanford.edu/papers/p...

@berak Yeah I went through that paper casually. We don't have to use permutohedral I guess, and a joint filter will work right?

hehe, i guess, i'll need another day to understand it ..

@berak. That rolling guidance paper is simple (a clever hack) rather. Permutohedral Lattice although might take some time. But I am not able to understand why they used that when in the paper they mentioned using Joint Bilateral Filter.

This thread has gone cold, but I'm wondering how this turned out. I did notice that ximgproc now includes 'rollingguidancefilter' and some other interesting edge-preserving filters, but I am apprehensive about rebuilding OpenCV. Is there any practical way to compile just the code for an individual filter within the ximgproc module, or is that module pretty monolithic?