Medical Image Fusion Based On Two-Level Multiscale Decomposition

The advent of medical imaging has improved the accuracy and convenience of clinical diagnosis.However,the image obtained by a single sensor has limitations,and obtaining a more complete image by complementing multiple sensors has become the current research direction of medical diagnosis.Aiming at the problems of insufficient image information extraction ability and poor image feature integration effect in existing multimodal medical image fusion algorithms,this paper proposes two new effective multimodal medical image fusion methods.The details are as follows:1.Aiming at the problem that the classification of image information is not precise enough in the existing image fusion literature,which leads to insufficient processing ability of image details,this paper proposes a two-level multi-scale decomposition medical image fusion algorithm based on the visual characteristics of the human eye.The algorithm uses NSST combined with NSCT to decompose source images at secondary multi-scale to obtain hierarchically richer image groups.Because the weighted average method used in traditional image fusion algorithms is easy to lose information,this paper proposes a fusion rule based on regional visual contrast,which uses an improved regional visual contrast algorithm to extract image feature information,which can reflect the correlation between images.to improve the accuracy of the algorithm for extracting feature information from the source image.In this paper,the effectiveness of the algorithm is verified from the subjective point of view and objective data through three sets of comparative experiments.2.In order to improve the effective information ratio of images,this paper proposes a two-level multi-scale medical image fusion algorithm based on saliency detection.The algorithm uses NSCT combined with the implicit low-rank algorithm to decompose the source image at two levels and multiple scales to obtain image layers with richer layers,and uses the implicit low-rank algorithm to denoise the image to a certain extent.In this paper,a fusion rule based on saliency detection is proposed,and the fusion image is obtained by calculating the superpixel distance as the input parameter of the simplified PCNN.After comparing with the simulation experiments of five traditional algorithms,it is confirmed that the algorithm proposed in this paper has the best overall performance.