A cascade configuration of the edge‐weighted image enhancement filter

Abstract

SUMMARYIn this paper, a new configuration of the edgeweighted high-pass filter and a design method for the weight function are proposed. A widely used image enhancement method is to amplify the high-frequency components of the image and to intensify the brightness differential of the edges. This method, generally called the unsharp masking method, allows enhancement effects to be obtained with a simple structure. If a linear high-pass filter is used as a method of extracting the high-frequency components from the input image, the noise components superimposed on the image appear in the high-pass filter output and are enhanced. In order to resolve this problem, a weight function defined from the edge characteristics is multiplied by the high-pass filter so that the amplitude of the high-pass filter output is suppressed in the flat image region. However, in the conventional edge-weighted high-pass filter, the noise components generated on the edge coordinates are strongly emphasized. Therefore, the edge coordinates of the output image fluctuate irregularly and discontinuities appear in the frame lines. In this investigation, in order to suppress the noise components on the edge coordinates, a new edge-weighted high-pass filter structure is proposed. In this structure, the high-pass filter is decomposed into cascaded connections. The weighting function is multiplied by the filter output from the previous stage. Further, in order to suppress noise amplification, the weight function is defined as a linear sum of the edge characteristic measures. By statistical analysis of the noise amplification, the parameters of the weight function minimizing the output noise variance are obtained. The proposed image enhancement method is experimentally compared with the conventional method and its effectiveness is confirmed.