Abstract
This paper proposes a wavelet based multilevel edge detection method that exploites spline dyadic wavelets and a frame work similar that of Canny's edge detector [2]. Using the recently proposed dyadic lifting schemes by Turghun-jan et al [1], spline dyadic wavelet filters have been constructed, which are characterized by higher order of regularity and have the potential of better inherent noise filtering and detection results. Edges are determined as the local maxima in the subbands at different scales of the dyadic wavelet transform. Comparison reveals that our method performs better than Mallat's and Canny's edge detectors.