Multi-modality Medical Image Fusion Algorithms Based On NSST

With the development of image fusion technology,the defects of medical images in different body positions are gradually compensated in a single mode,and the visual effect of medical images is significantly improved,which has great theoretical significance and practical value for clinical medical diagnosis.In recent years,the importance of fusion technology in the field of clinical medicine is increasing day by day.More and more new technologies are applied in this field,which promotes the development of fusion technology to a new stage.In view of the complexity of medical images and the different characteristics of low and high frequency sub-bands after multi-scale decomposition.In this paper,we decide to analyze the fusion algorithm in multi-scale domain.The main analytical task is as follows:1)In this paper,the imaging principle and characteristics of common medical images are analyzed in detail,and the advantages and disadvantages of multi-scale decomposition tools such as wavelet transform,contourlet transform and shearlet transform in image application are compared.In order to make up for the deficiency of single modal image in clinical treatment and improve the image fusion quality.Because of the advantages that the NSST transform has strong ability in extracting image detail features,and the effect is remarkable,a multi-modal medical image edge fusion algorithm based on NSST transform is proposed.Above all,the source image is decomposed into low-frequency subband and high-frequency coefficients by NSST decomposer.Secondly,the low-frequency coefficient of the image is processed by the region maximum difference comparison method to highlight the edge of the low-frequency information,while the improved edge energy method and edge intensity method are used to fuse the high-frequency areas with more detailed texture information.Finally,the fused low-frequency coefficient and high-frequency subband are transformed by NSST inverse to get the result image.2)Due to the variety and complexity of medical images,a new image fusion design is proposed based on careful analysis of the advantages of sparse transformation,considering the combination of sparse theory and NSST.Firstly,source image is decomposed into low-frequency subband and high-frequency coefficients by NSST decomposer.Secondly,in view of the poor sparsity of low-frequency,this paper decides to improve the poor sparsity oflow-frequency by using the sparse theory;Meanwhile,the high-frequency are processed by the relative standard deviation comparison method,and the final high-frequency coefficients fusion coefficients are determined according to the relative standard deviation and the energy value.Finally,the fused low-frequency coefficient and high-frequency subband are transformed by NSST inverse to get the result image.3)In this paper,experiments are carried out based on gray space and color space respectively,and relevant algorithms are selected for comparison.Then analyses the advantages and disadvantages of this algorithm from subjective and objective aspects.From the subjective point of view,this algorithm effectively improves the complementarity of information between multi-modal medical images.The richness of information in images is better,and the clarity of detail texture is higher.It is very helpful for clinical medical diagnosis.According to the objective index analysis,the algorithm in this paper performs very well on the edge strength,spatial frequency,standard deviation and other index values and the other index values also have good performance,which fully illustrates that the algorithm in this paper has strong ability in processing image structure information and edge information and retains the important information of the source image effectively.It reflects high practical value of clinical medicine.