Medical Image Fusion Based On Multi-Scale Transform And CS-PCNN

Medical image fusion is an emerging image processing method whose core consists of multi-scale transformation and high and low frequency fusion.It works by applying digital image processing techniques to extract the inherent features acquired by various imaging devices on the same target and fuse them into a clear,complete image.It is of great research value in health detection and disease treatment.The fusion of Computer Tomography(CT)and Magnetic Resonance Imaging(MRI)images has the advantages of high soft tissue definition,high density resolution and high spatial resolution,but fusion can lead to distortion and loss of edge information.To address these issues the following two aspects were studied:1)To tackle the unreasonable setting of the internal link matrix of the traditional Pulse Coupled Neural Network(PCNN)model,a new Area Ratio(AR)principle was proposed to optimize the PCNN.This method improved the correlation between the central pixel point and the neighbouring pixel points,sharpening the features of the target and enhancing the edge information of the image,resulting in a better visual effect of the PCNN-based fused image.2)To address the problems of the more complicated parameters and over-reliance on experience in the configuration of the PCNN algorithm,an improved adaptive system was proposed to enable the parameters to have self-regulation capabilities.According to this approach,the PCNN can control the number of iterations of neurons based on the input gray value,so that it can influence the frequency of firing pulses.At the same time,it can retain richer contour information,reduce image distortion and improve resolution.Both of these improvements to PCNN were implemented in the multi-scale transform domain and reduced the computational effort of PCNN by using the Compressed Sensing(CS)technique.It provides a new approach to medical diagnosis whilst giving new ideas for the fusion of remote sensing images and the fusion of infrared and visible images.Through comparative experiments,the feasibility and superiority of the new method was demonstrated.Figure 32;Table 6;Reference 48...