Mathematics Colloquia and Seminars

Return to Colloquia & Seminar listing

Segmentation in image analysis is the problem separating `objects' from their `background' in a given image. Usually, one starts with `edge detectors' that produce `edgels' which give clues for the locations of the objects boundaries. Classical edge detectors are the Marr-Hildreth, and Haralick or Canny edge detectors. Next, usually one should integrate these edgles into meaningful contours that indicate the boundaries of the objects. In this talk I'll review the classics, and then introduce a framework that allows us to give a two--dimensional variational explanation for the Marr-Hildreth and the Haralick-Canny like edge detectors. Based on these observations, an improved `active contour model' is suggested, and its performances are shown to be better than classical active contours when directional information about the edge location is provided. We present a general model that incorporates alignment as part of other driving forces of an active contour, together with the `geodesic active contour' model for regularization, and the minimal variance criterion.