Ideas:
– Use super pixel as a regularisation strategy by using voting of the probabilities to decide if the super pixel is tool or not. Combine this approach with clustering the super pixels by using RCD and GMM or SMM. The way to merge the two could be boosting or random forests.
– Calculate the RCD of each super pixel and then cluster them in two groups using GMM or SMM build a graph using the distance between the RCDs. Then use graph cuts for spatial regularisation.
– Cluster super pixels using the RCD. See how the result looks.
Meeting minutes, Luis and Tom, 7 Sep 2015