Adaptive edge detecting approach based on scale-space theory

Abstract

The main problem existing in many edge detecting approaches is that they are sensitive to the noise. There is a conflict between the precision of edge detection and the effect of the noise removal. The scale-space theory on the basis of the optimal edge detecting theory is introduced in this paper. Each specific edge corresponds to an optimal filter scale (OFS) and the value of the OFS of each edge point is calculated based on the optimal edge detecting theory. An automatic threshold is introduced in the process of edge detection in order to eliminate the ragged edge further. Based on the optimal scale and the automatic threshold, a fast adaptive edge detecting algorithm is proposed. Experiment results demonstrate that this algorithm can get either the noise removing effect of low-resolution filter or the edge detecting precision of high-resolution filter and make a better compromise between the precision of edge detection and the effect of the noise removal.