Perceptual impact of edge sharpness in images

Abstract

A method based upon subjective viewing tests to evaluate the perceptual impact of different extents of edge sharpness is presented and the most eye-pleasing sharpness (MEPS) for an image edge-sharpening process is derived. The findings with Laplacian of Gaussian edge-enhancement filter show that the baseline MEPS is about 2.6 times that of the local just-noticeable difference, and the actual MEPS is also dependent on the contrast increase in the surrounding areas. The proposed methodology can be used to determine the MEPS for a particular edge-enhancement process, and the resultant formulation for the perceptual impact of edge sharpness can be used for reference in control of edge enhancement, image reconstruction and de-blurring processing, as well as objective visual quality gauge.