|
A new nonrigid registration algorithm is proposed in this paper, which applies Markov-Gibbs random fields model (MGRF) to the nonrigid registration field of medical images. The algorithm is constructed by integrating the nonrigid registration algorithm based on B-spline wavelet and the priori knowledge of the maximum likelihood (ML) and maximum a posteriori (MAP) estimation into a MGRF model. In the MGRF model, the reference and test images are the known conditions and B-spline wavelet is the basis function constructing the nonrigid deformation function. The coefficients of B-spline wavelet in the deformation function are the parameters to be evaluated. The algorithm leads to a better result of registration because the use of the priori knowledge. Various medical images are selected to verify the algorithm, which show that the algorithm proposed in this paper is superior to the nonrigid registration algorithm without the priori knowledge. |
|
Keywords:Markov-Gibbs Random Fields, Nonrigid Registration, B-spline wavelet, priori knowledge |
|