Affordable Access

deepdyve-link
Publisher Website

Color code identification in coded structured light.

Authors
  • Zhang, Xu
  • Li, Youfu
  • Zhu, Limin
Type
Published Article
Journal
Applied Optics
Publisher
The Optical Society
Publication Date
Aug 01, 2012
Volume
51
Issue
22
Pages
5340–5356
Identifiers
DOI: 10.1364/AO.51.005340
PMID: 22859022
Source
Medline
License
Unknown

Abstract

Color code is widely employed in coded structured light to reconstruct the three-dimensional shape of objects. Before determining the correspondence, a very important step is to identify the color code. Until now, the lack of an effective evaluation standard has hindered the progress in this unsupervised classification. In this paper, we propose a framework based on the benchmark to explore the new frontier. Two basic facets of the color code identification are discussed, including color feature selection and clustering algorithm design. First, we adopt analysis methods to evaluate the performance of different color features, and the order of these color features in the discriminating power is concluded after a large number of experiments. Second, in order to overcome the drawback of K-means, a decision-directed method is introduced to find the initial centroids. Quantitative comparisons affirm that our method is robust with high accuracy, and it can find or closely approach the global peak.

Report this publication

Statistics

Seen <100 times