Skip to main content
. Author manuscript; available in PMC: 2010 Jul 6.
Published in final edited form as: J Proteome Res. 2009 Jul;8(7):3558–3567. doi: 10.1021/pr900253y

Figure 2.

Figure 2

Markov Random Field (MRF) model. Each node represents a pixel in the classification map, the optimal label assignment depends on both single site potential (φ) and pair potential (ϑ). The former is defined by the Random Forest output that assigns probabilities to the T class labels whereas Potts potentials are used for the latter.