Research On Non-rigid Image Registration Methods Based On Active Demons Algorithm

Active Demons algorithm is an important non-rigid registration method. Since only relying on homogeneous coefficient a adjusting deformation force strength during the iterative process, the Active Demons algorithm can’t handle the registration of large deformation and small deformation at one time.In order to solve this problem, we propose an improved non-rigid registration method based on Active Demons algorithm, in which a new parameter called balance coefficient k is introduced to the calculation of demons force. The balance coefficient k can adjust the force strength combined with homogeneous coefficient a. Thus, the adjustable deformation range will become bigger, and more flexible, and the new method can deal with both large deformation and small deformation.Moreover, in order to further improve the registration accuracy and the convergence speed and avoid falling into local extreme value, the multi-resolution strategy is adopted in the registration process. Experiments on synthetic, natural and medical images demonstrate that the proposed method is faster and more accurate than the classical Active Demons algorithm, and the registration accuracy is improved by 54.40% on average.