Real-time line matching from stereo images using a nonparametric transform of spatial relations and texture information

Jonghee Park; Kuk-Jin Yoon

doi:10.1117/1.OE.54.2.023106

19 February 2015 Real-time line matching from stereo images using a nonparametric transform of spatial relations and texture information

Jonghee Park, Kuk-Jin Yoon

Author Affiliations +

Optical Engineering, Vol. 54, Issue 2, 023106 (February 2015). https://doi.org/10.1117/1.OE.54.2.023106

Abstract

We propose a real-time line matching method for stereo systems. To achieve real-time performance while retaining a high level of matching precision, we first propose a nonparametric transform to represent the spatial relations between neighboring lines and nearby textures as a binary stream. Since the length of a line can vary across images, the matching costs between lines are computed within an overlap area (OA) based on the binary stream. The OA is determined for each line pair by employing the properties of a rectified image pair. Finally, the line correspondence is determined using a winner-takes-all method with a left-right consistency check. To reduce the computational time requirements further, we filter out unreliable matching candidates in advance based on their rectification properties. The performance of the proposed method was compared with state-of-the-art methods in terms of the computational time, matching precision, and recall. The proposed method required 47 ms to match lines from an image pair in the KITTI dataset with an average precision of 95%. We also verified the proposed method under image blur, illumination variation, and viewpoint changes.

Citation Download Citation

Jonghee Park and Kuk-Jin Yoon "Real-time line matching from stereo images using a nonparametric transform of spatial relations and texture information," Optical Engineering 54(2), 023106 (19 February 2015). https://doi.org/10.1117/1.OE.54.2.023106

Published: 19 February 2015

ACCESS THE FULL ARTICLE

INSTITUTIONAL
Select your institution to access the SPIE Digital Library.

SELECT YOUR INSTITUTION

PERSONAL
Sign in with your SPIE account to access your personal subscriptions or to use specific features such as save to my library, sign up for alerts, save searches, etc.

PERSONAL SIGN IN

No SPIE Account? Create one

PURCHASE THIS CONTENT

SUBSCRIBE TO DIGITAL LIBRARY

50 downloads per 1-year subscription

Members: $195

Non-members: $335 ADD TO CART

25 downloads per 1 - year subscription

Members: $145

Non-members: $250 ADD TO CART

PURCHASE SINGLE ARTICLE

Includes PDF, HTML & Video, when available