The used two algorithms for the urbanThe used two algorithms for the urban

The MRF belongs to a branch of probabilistic theory, which embodies a
strong relationship with physical phenomena. The pixel interaction in an image
has a spatial correlation, hence MRF analyses the image effectively  through texture statistical characteristics
by the conditional probability distribution function. MRF is widely applied to
edge detection, image segmentation, restoration and reconstruction and so on
48 . The first algorithm to propose a linear road  extraction from a RS image was created in 1998
39. Initially, the road candidate segments were extracted with a linear detector
in its local area. Subsequently, the real road segments were selected and
connected based on the MRF. In the later years, 44 a novel combined method
was proposed: Markov random texture model and the SVM classifier. This
semiautomatic road extraction method was primarily used for the “synonyms spectrum”
phenomenon. Although is utilized texture features for training , a need for
human intervention probed enhancements. Eventually a road extraction method was
established based on a MRF and hybrid model of the SVM and Fuzzy C-Mean (FCM)
54 . In the paper, the author used two algorithms for the urban RS images:
the MRF and hybrid model of the SVM and the FCM.