On one hand, it used the MAP-MRF framework to adapt the sampler
training and to get the factor of a model. The accuracy of road extraction was
85.43%. On the other hand, the SVM plus the FCM model was studied and
integrated together. Firstly, an image was clustered by the FCM. The clustering
accuracy increased by sampling accuracy. Secondly, the clustering result was
classified by the SVM. The accuracy of road extraction was 94.57%. Fig 4.

presents the two algorithm results. From the above analysis it can be seen that
the complex relationship among the extraction objects in a RS image makes it
difficult to be accurately modelled by only using the MRF model. Hence, the
hybrid probability model is expected to achieve good results in complex object
detection from RS images 

We Will Write a Custom Essay Specifically
For You For Only $13.90/page!

order now