Edge detection: problems and solutions

This paper provides a comprehensive discussion on the processes involved in an edge detection system. The review details the fundamental concepts of edge detection, from the choice of the detection method to the differentiation and to the reason why a particular derivative is used. The problem of noise is addressed, as is the appropriate method of filtering an image. The combination of these choices and their inherent effect on the compromise between the accurate localisation of the edges, and the immunity to the noise in the image is presented. The paper also outlines the reasons for the choice of the structure of the hybrid edge detector proposed previously by the author et al. (1997), and the adaptive and optimal selection of its parameters to attain both immunity to noise, and accurate localization of the edges.