The Radon transform and its application to shape parametrization in machine vision

摘要

A general method is presented that uses the Radon transform as a means of defining a two-dimensional transform space in which information about different, analytically defined shape primitives in an edge image space may be encoded simultaneously. Examples are given illustrating how the shape-indicative distributions within the transform space may be deduced. The results show that each set of coded information is transparent to any other and that each shape-indicative distribution may be located using a convolution mask peculiar to that distribution.