Medical Image Registration Based On Mutual Information Measure And Multi Resolution

In medicine,often for the fusion of images,and the image registration is necessary before fusion.As the important basis of the fusion,image registration is the key technology in medical image processing.Medical diagnosis has benefited from the 30 years development of image registration.Registration is very significant in medical image processing.At the same time,it also plays an important role in improving the medical level.The registration process has a number of interlocking links and hands the fit is very important.At present,the researchers are conducting a number of explorations for each step of medical image registration.Without the need for feature extraction and image preprocessing,gray medical image registration based on mutual information has attracted extensive attention.Many scholars have been actively devoted efforts into the study.More and more new medical image registration algorithms have been implemented in real life,the results achieved good accuracy and robustness.Image registration has gradually become one of the hotspots in the field of medical image processingIn this paper,we make some improvements on the basis of mutual information and multi-resolution strategy.The work done in this paper is as follows:(1)The paper introduces the research background and significance of medical image registration,the concept and the basic framework of the registration,and describes the image interpolation,mutual information measure and multi-resolution strategy.(2)In the process of image registration,the particle swarm optimization(PSO)algorithm is easy to fall into the local extreme value.Based on the multi-resolution strategy,an improved particle swarm optimization(PSO)algorithm is proposed.Self-regulation strategy and adaptive weights are introduced into the improved particle swarm algorithm,combined with wavelet transform to decompose the image formation image of Pyramid,using the improved particle swarm optimization in each layer image.The results of the previous image registration as a parameter of the particle swarm algorithm in the next image registration,in order to raise the accuracy of registration results.Through the registration of brain MRI and brain CT,brain CT images and their own transformation images,the registration experiment can achieve good results,and verify the effectiveness of the algorithm.(3)In order to solve the problem of poor convergence ability of Powell algorithm in image registration,an improved difference algorithm combine with Powell algorithm are proposed.At first,the source images are decomposed into three layers based on wavelet transform.Each layer using different optimization algorithm for registration: the first layer using the improved differential algorithm,the latter two layers using the Powell algorithm.An improved adaptive operator is introduced to solve the problem that the Powell algorithm is too dependent on the initial point.The wavelet transform based on multi-resolution method can speed up the process of image registration,and can also get good registration results.Through the analysis of the experimental results,based on mutual information and multi resolution strategy,the proposed algorithm can improve the accuracy of medical image registration,reduce the registration time and achieve good registration results.