Presentation 2012-03-29
A Study on performance improvement of SIFT detector using laplacian filter
Tosiaki INOUE, Mutsumi OHTA,
PDF Download Page PDF download Page Link
Abstract(in Japanese) (See Japanese page)
Abstract(in English) As rapid spread of various devices with built-in cameras, it is strongly expected to develop advanced image recognition technologies. Robust feature extraction methods such as SIFT and SURF are widely used in many kinds of generic object recognition tasks, and further improvements on these methods are essential to boost recognition performance up effectively. In this paper, we describe an improvement of SIFT on keypoint detection. In original SIFT, DoG filter instead of LoG function is used to detect keypoints from gaussian scale-space images, whereas in our simple method, fixed window laplacian filter is adopted to improve both approximation error and repeatability. Repeatability test results using benchmark datasets show that our proposed method outperforms original SIFT and SURF. Image-verification test results using ALOI datasets also show that the method is effective to improve discriminative performance.
Keyword(in Japanese) (See Japanese page)
Keyword(in English) SIFT / DoG / LoG / laplacian filter / repeatability
Paper # PRMU2011-241,HIP2011-69
Date of Issue

Conference Information
Committee HIP
Conference Date 2012/3/22(1days)
Place (in Japanese) (See Japanese page)
Place (in English)
Topics (in Japanese) (See Japanese page)
Topics (in English)
Chair
Vice Chair
Secretary
Assistant

Paper Information
Registration To Human Information Processing (HIP)
Language JPN
Title (in Japanese) (See Japanese page)
Sub Title (in Japanese) (See Japanese page)
Title (in English) A Study on performance improvement of SIFT detector using laplacian filter
Sub Title (in English)
Keyword(1) SIFT
Keyword(2) DoG
Keyword(3) LoG
Keyword(4) laplacian filter
Keyword(5) repeatability
1st Author's Name Tosiaki INOUE
1st Author's Affiliation Pioneer Corporation()
2nd Author's Name Mutsumi OHTA
2nd Author's Affiliation Pioneer Corporation
Date 2012-03-29
Paper # PRMU2011-241,HIP2011-69
Volume (vol) vol.111
Number (no) 500
Page pp.pp.-
#Pages 6
Date of Issue