To detect the centers of interesting regions from the multi-scale edge maps described previously, it is necessary to compute the degree of symmetric enclosure at each point in the image for each scale. Sela  proposes that symmetric enclosure can be computed from the axes of symmetry of the contours in an edge map. An interest map (or level of perceptual enclosure) can be determined from the intersection of lines of symmetry. Other methods exist for determining the level of perceptual enclosure such as Reisfeld's Symmetry Transform . However, Sela's technique  is particularly computationally efficient. The algorithm is described below and its discrete implementation is then outlined to demonstrate its efficiency.