Edge Detection Within an Annular Sampling Framework

In previous reports, annular sampling has been introduced as a means for detecting symmetrically enclosing edge conngurations. Starting with a single grey-scale image, before employing annular operators, it is necessary to rst extract edges from the image. In this report, we examine why it is necessary to extract edge at multiple scales when employing annular operators. The scale of edge detection is deened in terms of the spatial extent over which an edge has support in an image. Curvature constraints imposed by the annular sampling scheme are then used to deene an edge detection scheme in which the spatial extent of an edge is set in proportion to the annular scale. Results are demonstrated for several images acquired in natural settings.