Gaussian mixed model in support of semiglobal matching leveraged by ground control points

Hao Ma; Shunyi Zheng; Chang Li; Yingsong Li; Li Gui

doi:10.1117/1.JRS.11.025014

23 June 2017 Gaussian mixed model in support of semiglobal matching leveraged by ground control points

Hao Ma, Shunyi Zheng, Chang Li, Yingsong Li, Li Gui

Author Affiliations +

Journal of Applied Remote Sensing, Vol. 11, Issue 2, 025014 (June 2017). https://doi.org/10.1117/1.JRS.11.025014

Abstract

Semiglobal matching (SGM) has been widely applied in large aerial images because of its good tradeoff between complexity and robustness. The concept of ground control points (GCPs) is adopted to make SGM more robust. We model the effect of GCPs as two data terms for stereo matching between high-resolution aerial epipolar images in an iterative scheme. One term based on GCPs is formulated by Gaussian mixture model, which strengths the relation between GCPs and the pixels to be estimated and encodes some degree of consistency between them with respect to disparity values. Another term depends on pixel-wise confidence, and we further design a confidence updating equation based on three rules. With this confidence-based term, the assignment of disparity can be heuristically selected among disparity search ranges during the iteration process. Several iterations are sufficient to bring out satisfactory results according to our experiments. Experimental results validate that the proposed method outperforms surface reconstruction, which is a representative variant of SGM and behaves excellently on aerial images.

Citation Download Citation

Hao Ma, Shunyi Zheng, Chang Li, Yingsong Li, and Li Gui "Gaussian mixed model in support of semiglobal matching leveraged by ground control points," Journal of Applied Remote Sensing 11(2), 025014 (23 June 2017). https://doi.org/10.1117/1.JRS.11.025014

Received: 16 March 2017; Accepted: 7 June 2017; Published: 23 June 2017

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