Regresar

A wavelet transform based multi-resolution approach to edge profile detection and classification

Abstract:

In this paper a new multi-resolution method to detect and classify edges appearing in images has been proposed. The edge detection and classification schema is based on the analysis of the data obtained by a multi-resolution image analysis using Mallat and Zhong's wavelet. Eight geometrical contour types have been classified by simple numerical analysis, together with the 'edge noise' considered as a real edge type. The classification method is based on a polynomial-fitting algorithm, by using second and third order polynomials. We can filter out the noise simply by eliminating the so classified edge. The proposed classification schema could be generalized to real edge types: shadows, corners, etc.