Polygonal approximation of digital curves based on the principles of perceptual organization

A new approach to the problem of polygonal approximation of digital shapes is presented in this paper. In this approach the approximation process is divided into three stages based on the principles of perceptual organization, and three algorithms are developed accordingly. The Linking-Merging Algorithm is firstly proposed for low level processing. The Smoothing Algorithm is then followed for further approximation. The high level Rule-based Algorithm is finally applied to preserve the visual feature points while reducing the number of segments in the approximated polygon as much as possible. The effectiveness of the proposed algorithm is demonstrated by testing both synthetic shapes and scanned-in shapes. © 1997 Pattern Recognition Society. Published by Elsevier Science Ltd.