CURVED 3D OBJECT RECOGNITION USING 2D SHAPE ANALYSIS TECHNIQUES.

Wen-Hsiang Tsai *, J. C. Lin

*Corresponding author for this work

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

2 Scopus citations

Abstract

A 3-D object recognition system capable of measuring 3-D object surfaces is described. The system is simple, efficient and relatively inexpensive. It relies on parallel-stripe projections for 3-D surface measurement. Data from four camera views are collected to form reference models which consist of horizontal cross section boundaries. The recognition scheme engages 3-D object registration followed by 3-D similarity measurement. The 3-D registration is accomplished by multiple 2-D registrations on the cross section planes using the generalized 2-D Hough transform. The 3-D similarity measure uses the average of all the 2-D cross-sectional similarity measures which are based on the 'distance-weight correlation. ' Since the recognition scheme involves only pixel-level operations, it therefore is amenable to parallel processing. 26 refs.

Original languageEnglish
Title of host publicationUnknown Host Publication Title
PublisherIFS (Publications) Ltd
Pages379-393
Number of pages15
ISBN (Print)0903608758
StatePublished - 1 Dec 1984

Fingerprint

Dive into the research topics of 'CURVED 3D OBJECT RECOGNITION USING 2D SHAPE ANALYSIS TECHNIQUES.'. Together they form a unique fingerprint.

Cite this