A New Hardware Module for Stereo Matching Using Zernike Moments

Author

Gonidis, Paschalis ; Kotoulas, Leonidas ; Andreadis, Ioannis

Author_Institution

Democritus Univ. of Thrace, Thrace

fYear

2007

fDate

19-25 June 2007

Firstpage

33

Lastpage

33

Abstract

This paper presents a new hardware module suitable for matching point features between two uncalibrated images of a scene with real-time response. A robust matching is established under a mixed geometric and moment-based correspondence strength function. In order to provide rotation invariance and low noise sensitivity, Zernike moments are utilized. The proposed system is capable of extracting correspondences between stereo pairs at a rate of over 25 pairs per second, for VGA images.

Keywords

Zernike polynomials; computer vision; edge detection; feature extraction; image matching; pipeline processing; singular value decomposition; stereo image processing; SVD computation; VGA images; Zernike moments; computer vision; corner detector; geometric function; hardware accumulation grid architecture; low noise sensitivity; moment-based correspondence strength function; optical scene; point feature matching; real-time response; rotation invariance; stereo matching; uncalibrated images; Application software; Computer vision; Detectors; Hardware; Image edge detection; Image matching; Matrix decomposition; Robustness; Singular value decomposition; Stereo vision;

fLanguage

English

Publisher

ieee

Conference_Titel

Autonomic and Autonomous Systems, 2007. ICAS07. Third International Conference on

Conference_Location

Athens

Print_ISBN

978-0-7695-2859-7

Electronic_ISBN

978-0-7695-2859-7

Type

conf

DOI

10.1109/CONIELECOMP.2007.47

Filename

4437910