Multiscale extension of the gravitational approach to edge detection

Abstract

The multiscale techniques for edge detection aim to combine the advantages of small and large scale methods, usually by blending their results. In this work we introduce a method for the multiscale extension of the Gravitational Edge Detector based on a t-norm T. We smoothen the image with a Gaussian filter at different scales then perform inter-scale edge tracking. Results are included illustrating the improvements resulting from the application of the multiscale approach in both a quantitative and a qualitative way. © 2011 Springer-Verlag.