ENME Webcast Archives
LEADERS IN MECHANICAL ENGINEERING LECTURE SERIES - FALL 2007
Movie doesn't work?
Quicktime 7 is now
required to view Webcasts - and it's free.
Edge Detection using Separation and Decomposition of Scales
Lecturer: Eitan Tadmor - Department of Mathematics, CSCAMM and IPST - University of Maryland
Original Air Date: Friday, November 9th at 11:00am
Abstract: I shall describe two, not unrelated scenarios, where edges can be treated through different scales. The first part of the talk is devoted to detection of edges in piecewise smooth data. The data is given in terms of its spectral representation. The objective is to recover the location and amplitudes of jump discontinuities, from possibly incomplete and noisy spectral information. We utilize a general family of edge detectors based on concentration kernels. Each kernel forms an approximate derivative of the delta function, which detects edges by separation of scales.
Edges are also the noticeable feature in images. The second part of the talk is devoted to a novel representation of general images as hierarchy of edges, by decomposition of scales. The resulting decomposition is essentially nonlinear. The questions of convergence, localization and adaptivity will be discussed and numerical results will illustrate applications to synthetic and real images.
