Fast edge-based method for image magnification

For image magnification, nearest-neighbourhood interpolation, linear interpolation and bicubic interpolation are preferable for their rapidity, but they have a step-like boundary problem or blurring effect. To improve these defects, many edge-based methods have already been proposed, and some are being developed. Nevertheless, their high complexities prevent them from being popular. This paper proposes a new fast edge-based method. Several assumptions are made to speed up the edge estimation process, and several simple hybrid interpolations are proposed. Experimental results demonstrate that the proposed method is fast and the magnified images are vivid with sharp edges and smooth homogeneous areas.