Inverse distance weighted interpolation based on image topology for super-resolution

Super-resolution is the process of producing a high resolution image by using a series of slightly different low resolution images. Super-resolution, generally, is composed of three stages: registration of low resolution images for determining the distances between them, interpolation or merging algorithms, and image enhancement steps. In the second stage, which is the most important one, pixels of low resolution images are combined with weights to compute high resolution image pixels. Blurring of edges and corners, of which their preservations are important for image understanding, is one of the most important problems to deal with for these algorithms. This thesis is about the enhancement of algorithms used in the second stage by integrating connectivity and connected relationships of pixels into computations. In this work, an algorithm to find connectedness diagrams for non-uniform data and the effect of connectedness information to edges on the preregistered synthetic images is shown. Among different interpolation methods, Inverse Distance Weight (IDW) algorithm is preferred and high resolution image pixels are computed by integrating connectedness information into this algorithm. Results, obtained from the newly developed algorithm are compared to cubic interpolation and conventional IDW methods and it is shown that the new algorithm yields a superior performance in edge preservation and noise suppression compared to others.