Automatic edge extraction using locally adaptive threshold

摘要

The authors describe an automatic edge extraction algorithm using a new edge operator called the E/sup K/ edge operator, and a locally adaptive threshold subject to brightness and gradient value in a picture. The proposed algorithm is capable of extracting one-pixel-thick edges and giving a rather pleasing representation of the original grey-level picture. This algorithm is compared to the well known Sobel gradient operator and Keskes' zero-crossing operator.