Share Email Print

Proceedings Paper

How to avoid multiplications when generating Euclidean distance maps
Author(s): Oleg G. Okun
Format Member Price Non-Member Price
PDF $17.00 $21.00

Paper Abstract

A new implementation of the Euclidean, ordered propagation distance transform is developed. Unlike traditional approaches, where multiplication operations contribute a significant cost to computation, we suggest to apply the city lock and chessboard distance transform to approximate the Euclidean distances.In fact, the city block transform is just performed, while the chessboard one is simulated based on the iteration number value. Though multiplications are not totally excluded, their amount is greatly reduced because a few pixel values are only corrected at each iteration of the algorithm by using the Euclidean distances as compared to other similar methods, where those distances are determined for each pixel. As a result, we obtain a faster method for generating the Euclidean distance maps, which is still accurate.

Paper Details

Date Published: 2 October 1998
PDF: 8 pages
Proc. SPIE 3454, Vision Geometry VII, (2 October 1998); doi: 10.1117/12.323261
Show Author Affiliations
Oleg G. Okun, Institute of Engineering Cybernetics (Finland)

Published in SPIE Proceedings Vol. 3454:
Vision Geometry VII
Robert A. Melter; Angela Y. Wu; Longin Jan Latecki, Editor(s)

© SPIE. Terms of Use
Back to Top