Region Edges

Abstract

Edges play a dominant role in computer vision research, but have received little attention so far. Although independent edge detection and edge linking algorithms are not of direct concern to the digital surface segmentation algorithm, the topic of parametric edge description is of significant interest. Since the approximate shape of smooth graph surfaces in images is given parametrically by the approximating graph surface functions z = ĝ _i (x, y), it would be convenient if the support region of the graph surface could also be formulated parametrically as (x(s), y(s)). However, the 2-D image regions associated with these approximants are specified in a non-parametric, digital form as a list of connected image pixels at this point in the algorithm. The boundary of an image region can be encoded very compactly as an 8-neighbor chain code, which is a general method for describing region boundaries in a discrete image, but it is not as desirable as a parametric curve representation.