The master’s thesis presents and evaluates a method for image magnification and reconstruction, the Generalized constrained interpolation method. It is based on interpolation, which is used for initial image magnification. In the next step it eliminates visual artifacts using reconstruction models. Three models are used: a contour smoothing model, an edge sharpening model, and a sensor model.
The method was developed with the goal of creating a high-quality general method for image enlargement with good user experience. The main advantage is that it can enlarge an arbitrary image by an arbitrary magnification factor. It falls within the time range of complex interpolation methods. The result of the method are high-quality images with smooth contours and clear edges, which can be compared to the results of good interpolation methods, such as the NEDI method.
|
