ENCODING AND DECODING WITH REFINEMENT OF THE RECONSTRUCTED PICTURE

An encoding method is disclosed. A picture part is encoded (S 120) in a bitstream and reconstructed comprising. Refinement data are determined (S 130) such that:- a rate distortion cost computed as a weighted sum of data coding cost and of a distortion between an original version of said picture part and said reconstructed picture part after refinement by said refinement data is decreased compared to a rate distortion cost computed as a weighted sum of data coding cost and of a distortion between said original version of said picture part and said reconstructed picture part without refinement in the case where no mapping applies before coding;- a rate distortion cost computed as a weighted sum of data coding cost and of a distortion between an original version of said picture part and said reconstructed picture part after inverse mapping and refinement by said refinement data is decreased compared to a rate distortion cost computed as a weighted sum of data coding cost and of a distortion between said original version of said picture part and said reconstructed picture part after inverse mapping without refinement in the case where mapping applies before coding and said refinement is out of the decoding loop;- a rate distortion cost computed as a weighted sum of data coding cost and of a distortion between a mapped original version of said picture part and said reconstructed picture part after refinement by said refinement data is decreased compared to a rate distortion cost computed as a weighted sum of data coding cost and of a distortion between said mapped original version of said picture part and said reconstructed picture part without refinement in the case where mapping applies before coding and said refinement is in the decoding loop.Finally, the refinement data are encoded (S 140) in the bitstream.