skip to primary navigation skip to content

CBSU bibliography search

To request a reprint of a CBSU publication, please click here to send us an email (reprints may not be available for all publications)

Representing the shape of image contours.
Watt, R.J. & Nimmo-Smith, I.
In J.C. Simon (Ed.), From Pixels to Features (pp. 333-344). Elsevier Science Publishers, B.V. (North-Holland).
Year of publication:
CBU number:
There is a multitude of different ways in which the two-dimensional shape of contours in an image could be represented. In this paper we discuss the computational theory of the strategy apparently adopted by human vision. This strategy has three parts. Initially, corners, intersections and line ends are identified and localized. The line segments that connect these points are guaranteed to be continuous and smooth (twice differentiable). The next step is to measure the mean curvature of each line segment, and finally any bumps on the curve are detected and analysed. These three steps are examined from two points of view: Firstly, we consider how the overall description that is created is related to three-dimensional structures in the scene; then secondly we discuss the strategy as an instance of signal processing where the signal is unknown.