A Fast Medical Image Registration Algorithms Based on Steerable Pyramid
Bo Hu · Fudan xuebao. Ziran Kexue ban · 2010
A fast algorithm for medical image registration is presented.With steerable filters,the multi-directional feature information of image can be steadily extracted and then the synthesized feature graph for more accurate registration can be obtained.To avoid heavy computation due to the globally searching rotation center,a method to determine the centroid of feature graph as rotation center for geometry transform is presented,which uses steerable pyramid for level by level registration.Furthermore,centroid calculation can also provide initial estimation of translation parameters,which greatly accelerates the search process.Simulation results prove that the accuracy of the algorithm is high and the computation time of the algorithm can be remarkably reduced.For multimodal image registration,the algorithm also leads to a good result.