Modelling of Learning and Perception

Centre of Excellence in Computational Complex Systems Research

 

Image Segmentation by MCMC Methods

Example results of our segmentation algorithm

Images shown here have been scaled down by 50 %. Computations were done on 321 x 481 images. Same parameter values were used for all images. Full resolution images as well as segmentation results by human subjects can be found in The Berkeley Segmentation Dataset website.

Original image Final sample of MCMC chain Average of MCMC chain
Original image Final sample of MCMC chain Average of MCMC chain
Original image Final sample of MCMC chain Average of MCMC chain
Original image Final sample of MCMC chain Average of MCMC chain
Original image Final sample of MCMC chain Average of MCMC chain
Original image Final sample of MCMC chain Average of MCMC chain
Original image Final sample of MCMC chain Average of MCMC chain
Original image Final sample of MCMC chain Average of MCMC chain
Original image Final sample of MCMC chain Average of MCMC chain
Original image Final sample of MCMC chain Average of MCMC chain
Original image Final sample of MCMC chain Average of MCMC chain
Original image Final sample of MCMC chain Average of MCMC chain
Original image Final sample of MCMC chain Average of MCMC chain
Original image Final sample of MCMC chain Average of MCMC chain
Original image Final sample of MCMC chain Average of MCMC chain
Original image Final sample of MCMC chain Average of MCMC chain
Original image Final sample of MCMC chain Average of MCMC chain
Original image Final sample of MCMC chain Average of MCMC chain
Original image Final sample of MCMC chain Average of MCMC chain
Original image Final sample of MCMC chain Average of MCMC chain
Original image Final sample of MCMC chain Average of MCMC chain

>>>>>>>>>>>>>>>>>>>>>> next page >>>>>>>>>>>>>>>>>>>>>>
0 1 2 3 4 5 6 7 8 9