how to implement Markov random field to remove miss classified objects

Hello; i have a binary image which is a result of kmean clustering method,

hence the change map still needs smoothing;

is it possible to implement Markov random field to reduce the miss classified objects?

the attached zip correspond to the binary map

[Mahmoud Afifi]

Please check this code. It segments the image using Markov random field after Kmeans clustering. I guess this code can be modified to do your task.

https://www.mathworks.com/matlabcentral/fileexchange/33592-image-segmentation-based-on-markov-random-fields