how to do a adjustment map for tonal stabilization for videos

ndk

Hello. I want to implement on Python an algorithm of tonal stabilization for video from the paper: http://www.cs.huji.ac.il/labs/cglab/projects/tonestab/files/TonalStabilization-final.pdf

But I don't fully understand the space color in which the author working. I understand, that I have to normalize the three RGB channel with the Luminance channel: L= R+G+B

           R=R/L
           G=G/L
           B=B/L

By doing so, the author separate between Luminance and Chroma. That's I think I get it. And here their idea, and what I don't understand:

they represent the change at each pixel as one offset for the log luminance, which is log(R+G+B) or log (R)+log(G) + log( B) ? and two offset for the chroma 2 chroma channels, which are R/L and G/L ? Can you also explain to me how this represents an exposure correction ? Moreover, to initialise the adjustment Map,they say that for the pixel which have a very few difference (say <0.05) in intensity between 2 frames, f1 and f2, the correction consist to add the corresponding difference f1-f2 to the map. Ok, but then how do I add that to the log luminance, and chroma channel ?

Sorry for my poor english and don't hesitate to ask me for more precision in my request

koogs

(You need to use two newlines between paragraphs)

ndk

Sorry for that, I try again:

Hello. I want to implement on Python an algorithm of tonal stabilization for video from the paper: http://www.cs.huji.ac.il/labs/cglab/projects/tonestab/files/TonalStabilization-final.pdf

But I don't fully understand the space color in which the author working. I understand, that I have to normalize the three RGB channel with the Luminance channel:

L= R+G+B R=R/L G=G/L B=B/L

By doing so, the author separate between Luminance and Chroma. That's I think I get it. And here their idea, and what I don't understand:

they represent the change at each pixel as one offset for the log luminance, which is:

log(R+G+B)

or

log (R)+log(G) + log( B)

?

and two offset for the chroma 2 chroma channels, which are:

R/L and G/L ?

Can you also explain to me how this represents an exposure correction ? Moreover, to initialise the adjustment Map,they say that for the pixels which have a very few difference (say <0.05) in intensity between 2 frames, f1 and f2, the correction consist to add the corresponding difference f1-f2 to the map. Ok, but then how do I add that to the log luminance, and chroma channel ?

Sorry for my poor english and don't hesitate to ask me for more precision in my request