Interpolation of tomographic data using a self-similar approach

In this paper, a new method for slice interpolation of a tomographic dataset is presented. The method exploits the self-similarity that exists in the dataset, namely a part of the dataset is a scaled down copy of another larger part of the set. According to the proposed method, the interpolated slice is created using information from the four nearest available slices. Initially, correspondences between points, belonging to a properly chosen pair of the available slices, are established. These correspondences are used to guide a procedure for creating a slice that is an approximation of the interpolated slice. The same procedure is applied for the other pair of slices resulting in another approximation of the interpolated slice. A weighted sum of the two approximations is used in order to create the final version of the interpolated slice. Quantitative and qualitative results stress the superior performance of the proposed method against other widely used interpolation methods.