Local Multi-Modal SIFT Features in Co-registered Range and Intensity Images

Author

Strachan, E. ; Siebert, J.P.

Author_Institution

Sch. of Comput. Sci., Univ. of Glasgow, Glasgow, UK

fYear

2012

fDate

25-29 Nov. 2012

Firstpage

426

Lastpage

432

Abstract

SIFT features have known limitations for matching features under changes in 3D pose of the object. We investigate the use of multimodal information from the range and intensity domains as an approach to mitigating the effects of changes in view point. In this investigation we outline an experimental procedure applied to range and intensity images of real-world objects captured using a stereo camera pair and calibrated turntable setup offering reliable ground truth for the 3D motion of surface locations. From our investigation we propose a novel surface descriptor, EC-SIFT, which shows a significant improvement on state of the art performance in the experiments we present.

Keywords

feature extraction; image matching; image motion analysis; stereo image processing; 3D motion; 3D pose; EC-SIFT; calibrated turntable setup; coregistered range images; feature matching; intensity images; local multimodal SIFT feature; stereo camera pair; surface descriptor; surface location; Cameras; Feature extraction; Indexes; Shape; Surface treatment; Transforms; 2.5D SIFT; 2.5D feature extraction; local features; range features; range image;

fLanguage

English

Publisher

ieee

Conference_Titel

Signal Image Technology and Internet Based Systems (SITIS), 2012 Eighth International Conference on

Conference_Location

Naples

Print_ISBN

978-1-4673-5152-2

Type

conf

DOI

10.1109/SITIS.2012.69

Filename

6395126