Image Enhancement Based on Edge Profile Acutance

Abstract

The sharpness of an image is related to the higher-frequency content of the image and to the edge information in the image. Increasing values of denvative measure. across edges should relate sharper and more clearly perceivable image details. The mean-squared gradient is a reliable measure of edge sharpness, and has heen used in the definition of the "acutance" of an edge or region of Interest (ROI), Acutance could also serve as a local measure of image quality or the perceptibility of a region or feature of interest in an image. We propose a new method of comparing image edge profile acutunce based on the mean-aquared gradient along the normals to the boundary of an ROI Image enhancement techniques arc then proposed based on the idea or increasing the acutance of an ROI. For this purpose one-dimensional operator are applied to sets of pixels along the normals at each boundary pixel of an ROI.

The acutance algorithm has been tested on different test images, and the resulting values have been found to relate well to the perceived sharpness of the image. The enhancement method has been tested with different blurred test images, and has been found to increase their sharpness as well as the objective measure of acutance.