The Research Of Medical Image Fusion Algorithms Using Multiscale Analysis And Pulse Coupled Neural Network

With the development of science and technology,image technology has made great progresses.As the single property of photography collected by single sensor,it is limited in the implementation of extracting,analyzing and the post processing contents of images.So,the research of image fusion comes up.Image fusion is a technique that fuses images from different sensors or with different properties by established methods into one image.Since it was invented,it has been increasingly popular and played an more and more important role in people’s work and life,especially the application of medical image fusion in career of clinical care and achieved great success.Medical image fusion is the process of fusing multimodal photographs such as MRI-PET,MRI-SPECT,MRI-CT into one image,which implemented by suitable methods.Medical image fusion is the important branch of image fusion,also plays a significant role in improving the rate of clinical diagnosis,accurate treatment of disease.The works of the paper are as follows:1)The paper summarized several major multi-scale analysis methods,and introduced the principle of the classic pulse coupled neural networks;summarized several current major evaluation metrics of image fusion;made summarized introduction on the usual fusion rules:2)Based on the traditional spatial frequency,for the fusion of MRI-SPECT and MRI-PET,a gradient energy calculation method based on pseudo-space frequency was proposed,which formed a high-frequency fusion rule with PCNN,and in the local Laplacian domain,the gradient and detail information of the fused image were improved;3)In order to integrate the detail information into the final image well,such as edge information,gradient information and so on,for the fusion of MRI-SPECT and MRI-PET,a medical image fusion algorithm based on two different high frequency fusion rules were proposed under Local Laplacian Filter;4)To improve the anti-noise ability of the image and accurately extract the saliency components of the image,for the fusion of MRI-CT,proposed a medical image fusion method based on Latent low-rank representation and NSST.Experiments demonstrated that,the information such as the ratio of signal to noise has been improved compared with the contrast algorithms.