Share Email Print

Proceedings Paper

Geometrical measures of the similarity of gray-scale images
Format Member Price Non-Member Price
PDF $14.40 $18.00

Paper Abstract

There are papers describing measures of correspondence or similarity between two binary images or their parts, but only two papers suggest a measure for a comparison of objects of two grey-scale images. However, there are numerous applications of a measure for grey-scale images as whole entities. A useful application is the comparison of different algorithms devoted to the same task (edge detection, thresholding, image enhancement, segmentation and image reconstruction). This paper proposes some results to define such a measure. They are based on two different representations of grey-scale images: as `surfaces' and as `stacks' or umbra. We study an adaptation of some known formulas used for binary images to grey-scale images, and present a geometrical variant of such a measurement. We study different measures of diversity, based on different digital metrics, direct calculations of distances, and digital functions adapted to grey-scale images. We show that the `stack' representation needs more calculation time and that measures based on the representation are not sensitive to small image shifts, but very sensitive to noise.

Paper Details

Date Published: 11 August 1995
PDF: 12 pages
Proc. SPIE 2573, Vision Geometry IV, (11 August 1995); doi: 10.1117/12.216415
Show Author Affiliations
Valery V. Starovoitov, Institute of Engineering Cybernetics (Belarus)

Published in SPIE Proceedings Vol. 2573:
Vision Geometry IV
Robert A. Melter; Angela Y. Wu; Fred L. Bookstein; William D. K. Green, Editor(s)

© SPIE. Terms of Use
Back to Top