Share Email Print

Proceedings Paper

A Comparison of Target Detection and Segmentation Techniques
Author(s): John A. Hird; David F. Wilson
Format Member Price Non-Member Price
PDF $17.00 $21.00

Paper Abstract

A wide variety of techniques has been examined in the literature for the detection and segmentation of target objects in images. This paper is concerned with the comparison of a set of alternatives drawn from two generic approaches to the problem. Histogram-based techniques focus on the distribution of some descriptive attribute or set of attributes within the image. The use of several such algorithms is considered including a sampled peak-finding method, a sampled percentile-finding method, multivariate histogramming based on greylevel and edge information and the well-known superspike algorithm. Hierarchical target detection techniques, on the other hand, attempt to exploit links between multiple reduced resolution views of the image. A range of such methods is also described based on the use of both iterative and top-down traversal procedures. Each of the algorithms is discussed, and their performance on a database of synthetic and real infra-red images is compared in terms of segmentation quality and computational cost.

Paper Details

Date Published: 1 April 1990
PDF: 12 pages
Proc. SPIE 1191, Optical Systems for Space and Defence, (1 April 1990); doi: 10.1117/12.969682
Show Author Affiliations
John A. Hird, Pilkington Optronics (United Kingdom)
David F. Wilson, Pilkington Optronics (United Kingdom)

Published in SPIE Proceedings Vol. 1191:
Optical Systems for Space and Defence
Alan H. Lettington, Editor(s)

© SPIE. Terms of Use
Back to Top
Sign in to read the full article
Create a free SPIE account to get access to
premium articles and original research
Forgot your username?