Damaged CT Image Restoration For Dental Application

Since Roentgen discovered the X-ray more than 100 years ago, the medical image has experienced the revolutionary development. The contemporary medical image researches fully display the characteristic of multiplex-knowledge research. The medical(especially anatomy,neurology) knowledge,the physical conception,the mathematics method and the computer programming thought have the organic fusion during the medical image processing research. The application of CT,MRI,PET and so on many kinds of medical image formation pattern, as well as the visualization of three dimensional medical image, specially the internal organs and the visualization of three dimensional volume determination of brain, already become the daily patient diagnosis and the constituent of standard nursing procedure. The automatic diagnosis and diagnosis of pathology image also needs the image restoration,division,the quantification,the enhancement tool, as well as the effective algorithm about visualization.In this paper, we mainly handle the problem of medical CT image restoration. From the real application, we make an improvement on the texture synthesis by patch-based sampling. In our application architecture, we need the image imaging,division and registration. The damaged CT images bring difficulties to the accuracy. So we must put forward the solution.The texture synthesis by patch-based sampling is the latest one, also is the hottest one. We use texture synthesis by patch-based sampling to solve the restoration. Because the dental CT image has the folloing disadvantages: low SNR,low contrast,structure alterable,borderline faintness between different soft structure,shape complexity, the traditional algorithm can not solve this problem well. In our realization, we design a new search path and block matching strategy. At the same time, we use GPU to accelerate the process. Up to now, there is no...