Share Email Print

Optical Engineering

Wavelet-based adaptive thresholding method for image segmentation
Author(s): Zikuan Chen; Yang Tao; Xin Chen; Carl Griffis
Format Member Price Non-Member Price
PDF $20.00 $25.00
cover GOOD NEWS! Your organization subscribes to the SPIE Digital Library. You may be able to download this paper for free. Check Access

Paper Abstract

A nonuniform background distribution may cause a global thresholding method to fail to segment objects. One solution is using a local thresholding method that adapts to local surroundings. In this paper, we propose a novel local thresholding method for image segmentation, using multiscale threshold functions obtained by wavelet synthesis with weighted detail coefficients. In particular, the coarse-to-fine synthesis with attenuated detail coefficients produces a threshold function corresponding to a high-frequency-reduced signal. This wavelet-based local thresholding method adapts to both local size and local surroundings, and its implementation can take advantage of the fast wavelet algorithm. We applied this technique to physical contaminant detection for poultry meat inspection using x-ray imaging. Experiments showed that inclusion objects in deboned poultry could be extracted at multiple resolutions despite their irregular sizes and uneven backgrounds.

Paper Details

Date Published: 1 May 2001
PDF: 7 pages
Opt. Eng. 40(5) doi: 10.1117/1.1360243
Published in: Optical Engineering Volume 40, Issue 5
Show Author Affiliations
Zikuan Chen, Univ. of Arkansas (United States)
Yang Tao, Univ. of Maryland (United States)
Xin Chen, Univ. of Maryland (United States)
Carl Griffis, The Univ. of Arkansas (United States)

© SPIE. Terms of Use
Back to Top