Parameter estimation in Markov random field based on evolutionary programming

It's very difficult to estimate the parameters in the Markov random field due to the computationally intractable partition function when employing the Markov random field as the prior model of the image. This paper presents a new method to estimate these parameters employing evolutionary programming to search for the suitable parameters so that the difference between the simulated image based on these estimated parameters and the original image reaches minimum. Using this new method, the calculation of the computationally intractable partition function can be avoided. Furthermore, the most similar simulated image to the original image can also be obtained, which makes this method better than other traditional methods based on the likelihood function. The feasibility of this method is verified by experimental results.