Multi-scale Morphologic Tracking Approach for Edge Detection

Abstract

Aimed at eliminating the noise in the images, an edge detection algorithm based on multi-scale morphologic edge tracking approach is introduced in this paper. While edge detecting it can restrain the noise efficiently, if proper threshold parameters are selected. First, the morphology gradient is calculated under each scale. Second, after tracing the edge images acquired by sequential scale structuring elements, noises have been removed and useful edge informations have been reserved. Third, different weights are imposed to construct the outcome edge image. By comparing with other edge detection algorithms, the feasibility and superiority of this algorithm have been verified according to the experiment result.