Normalized Weighting Schemes for Image Interpolation Algorithms

This work presents four weighting schemes based on geometric shapes for image interpolation purposes. Geometric shapes of interests include the circle and triangle. The quantity used to express the extent of each shape is the normalized area. In the first scheme, the minimum diameter (MD) of a tetragon is used. In the second, the hypotenuse radius (HR) is used. In the third, the virtual pixel length based height for the area of the triangle (AT) is used. In the fourth scheme, the virtual pixel length for hypotenuse-based radius for the area of the circle (AC) is used. Objective assessments demonstrated that, at a higher scaling ratio, the AC scored highest at 60 % among new schemes presented. Subjective assessments demonstrated that the quality of images interpolated by the AC algorithm was superior or comparable to that of images interpolated by MD, HR, AT and selected traditional algorithms.