Comparison of Edge Detection in Robinson and Kirsch Methods

Abstract

This study aims to compare Robinson's edge detection method and Kirsch's method on image with a focus on the disclosure of special features such as texture, watermark, and design elements. Robinson's edge detection method uses a series of filters with eight neighboring pixel operations, while Kirsch's method uses a series of filters with more specific filter orientation to produce sharper edge responses. Paper money images were selected as research objects because they had distinctive features relevant to edge detection, such as differences in intensity on edge lines, smooth paper textures, and special patterns on watermarks. This research using banknote image and lung X-rays image dataset. From the results of comparison of edge detection with Robinson's method and Kirsch method it can be concluded that on Robinson's method the edge image of banknotes displays more detailed design elements of banknotes such as hero photographs, watermark, logo, and nominal. In the Kirsch method the bank image has a sharp edge response so that many of the design elements on the banknote are not clearly visible and contrasted with other banknotes. In a comparison of edge detection between Robinson and Kirsch's methods on pneumonia-infected lung X-rays, it can be inferred that Robinson produced a fine edge line but was difficult to find infection, while Kirsch produced a rough edge line that clarified infection in pneumonia-infected lungs.