GPU-ASIFT: A fast fully affine-invariant feature extraction algorithm

Author

Codreanu, Valeriu ; Feng Dong ; Baoquan Liu ; Roerdink, Jos B. T. M. ; Williams, Doug ; Po Yang ; Yasar, Burhan

Author_Institution

Sci. Visualization & Comput. Graphics, Rijksuniv. Groningen, Groningen, Netherlands

fYear

2013

fDate

1-5 July 2013

Firstpage

474

Lastpage

481

Abstract

This paper presents a method that takes advantage of powerful graphics hardware to obtain fully affine-invariant image feature detection and matching. The chosen approach is the accurate, but also very computationally expensive, ASIFT algorithm. We have created a CUDA version of this algorithm that is up to 70 times faster than the original implementation, while keeping the algorithm´s accuracy close to that of ASIFT. It´s matching performance is therefore much better than that of other non-fully affine-invariant algorithms. Also, this approach was adapted to fit the multi-GPU paradigm in order to assess the acceleration potential from modern GPU clusters.

Keywords

affine transforms; feature extraction; graphics processing units; image matching; parallel architectures; pattern clustering; ASIFT; CUDA version; acceleration potential; affine-invariant scale-invariant feature transform; feature extraction algorithm; graphics hardware; image feature detection; image matching; modern clusters; multi-GPU paradigm; Accuracy; Feature extraction; Graphics processing units; Hardware; Instruction sets; Kernel; Mathematical model; Computer Vision; GPU; High performance computing;

fLanguage

English

Publisher

ieee

Conference_Titel

High Performance Computing and Simulation (HPCS), 2013 International Conference on

Conference_Location

Helsinki

Print_ISBN

978-1-4799-0836-3

Type

conf

DOI

10.1109/HPCSim.2013.6641456

Filename

6641456