Abstract: Medical images,e.g.computed tomography(CT),magnetic resonance imaging(MRI) and positron emission tomography (PET),have important significance during the process of diagnosis and treatment for a lot of diseases.However,influenced by the restriction of equipment resolution and radiation dosage,the low resolution problem of medi-cal images is likely to adversely affect the final diagnosis and treatment.Aiming at this problem,a medical image super-resolution reconstruction algorithm of adaptive patch clustering was proposed.Firstly,a set of image patches in different scales,which is adaptive access to gray consistency,can be obtained by using of the algorithm of quad-tree decomposition for images.Then,the algorithm extracts features of these image patches,and clusters the patches to many centers of different scales after the process of clustering.Finally,the different scale centers will be used to reconstruct a high resolution image according to the clustering centers and the corresponding regression coefficients.The experimental results show that the new method performs better in medical image reconstruction,peak signal-to-noise ratio (PSNR) and structural similarity (SSIM).

Key words: Medical image,Super resolution reconstruction of image,Quad-tree decomposition,Clustering,Self-adaption