Research On Some Key Problems Of Image Registration Algorithm

Image registration is an important branch of image processing.It has a wide range of applications in the field of video image analysis,remote image,medical motion image,and feature recovery of3D scene and so on.In recent years,the research work of rigid image registration has made a lot of progress,but the two key issues of non-rigid image registration and 3D image registration still need to be studied further.To solve the two problems,based on the optical flow registration model and the two conservation techniques,this paper develops two improved non-rigid image registration algorithms.Moreover,the proposed algorithm is applied to3D medical image processing.The main contents of this paper are as follows:1.Deformation driving force of traditional Active Demons algorithms is lack of anti-noise performance.To overcome this disadvantage,by using the conservation of pixel and conservation of connected region,we propose an improved Active Demons algorithm with multi-resolution strategies.Experiment results show that the proposed algorithm can effectively enhance the visual effect and evaluation index of image registration.2.Conventional Additive Demons algorithms in two-dimensional and three-dimensional image registration suffer from low image registration accuracy and long computational time.To overcome this disadvantage,we propose an improved two-dimensional and three-dimensional Additive Demons image registration algorithms based on the local standard deviation of image conservation term.Experimental results show that the proposed algorithm has better registration accuracy and faster running time than the conventional Additive Demons algorithms.3.The proposed registration algorithm is applied to lung medical image super-resolution reconstruction.The experimental results show that the proposed registration algorithm can be effectively and successfully applied to the field of lung image super-resolution reconstruction.