GPU-based rectification of high-resolution remote sensing stereo images

Niko Lukač; Borut Žalik

doi:10.1117/12.2066988

21 October 2014 GPU-based rectification of high-resolution remote sensing stereo images

Niko Lukač, Borut Žalik

Proceedings Volume 9247, High-Performance Computing in Remote Sensing IV; 92470P (2014) https://doi.org/10.1117/12.2066988
Event: SPIE Remote Sensing, 2014, Amsterdam, Netherlands

Abstract

One of the major challenges in the topographic mapping and fast generation of digital terrain or surface models from stereo optical aerial or satellite imagery is the stereo rectification preprocessing step. The general case is the use of extrinsic and intrinsic parameters from each calibrated camera, in order to establish epipolar geometry. Stereo rectification consists of geometric transformation and image sub-pixel resampling. Such a task is computationally demanding when dealing with high-resolution optical imagery. This presents an increasingly evident problem as the remote sensing technologies are becoming more accurate, causing even higher computational demands. This paper proposes a novel method for fast rectification of stereo images pairs to epipolar geometry by using General Purpose computing on Graphics Processing Units (GPGPU). The method is capable of resampling large high-resolution imagery on-the-fly, due to efficient out-of-core processing. In the experiments a runtime comparison was made between the proposed GPU-based and multi-core CPU methods over a dataset consisting of 420 stereo aerial images, where the proposed method achieved a significant speedup.

Citation Download Citation

Niko Lukač and Borut Žalik "GPU-based rectification of high-resolution remote sensing stereo images", Proc. SPIE 9247, High-Performance Computing in Remote Sensing IV, 92470P (21 October 2014); https://doi.org/10.1117/12.2066988

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