Edge detection of digital image with different edge types

Abstract

Abstract In digital images processing, there are three types of edges based on intensity changes. Namely, step edges, ramp edges and edges noise. An edge is defined as a set of pixels where there is an abrupt change in colour intensity over distance. On-ramp edges where gray levels change slowly, the Gradient Method is able to detect better. On step edges where the intensity or gray levels changes very quickly the Laplace method is able to detect better than the Gradient Method. In this study, three images were used as samples and identified the type of edge of each image. Furthermore, edge detection is performed with the first derivative operator Canny and the second derivative operator Laplacian of Gaussian. The results indicate that for step edges LoG provides better results, whereas for ramp edges Canny detects better. However, by selecting the right threshold that matches the σ (standard deviation), Canny is also capable to provide good edge detection results. The greater the σ value, the threshold was chosen must be small so that the results obtained are good and easily interpreted. The Canny operator produces a thinner edge and a firmer boundary between objects and between objects on the given sigma = 1 value while the LoG operator corrects better, especially on the steep part of the value σ = 2 compared to the value σ = 1.