Hardware-Efficient Local Extrema Detection for Scale-Space Extrema Detection in SIFT Algorithm

Kazuhito ITO  Hiroki HAYASHI  

Publication
IEICE TRANSACTIONS on Fundamentals of Electronics, Communications and Computer Sciences   Vol.E99-A   No.12   pp.2507-2510
Publication Date: 2016/12/01
Online ISSN: 1745-1337
DOI: 10.1587/transfun.E99.A.2507
Type of Manuscript: Special Section LETTER (Special Section on VLSI Design and CAD Algorithms)
Category: 
Keyword: 
SIFT,  local extrema detection,  scale-space extrema detection,  

Full Text: PDF(265.1KB)>>
Buy this Article




Summary: 
In this paper a hardware-efficient local extrema detection (LED) method used for scale-space extrema detection in the SIFT algorithm is proposed. By reformulating the reuse of the intermediate results in taking the local maximum and minimum, the necessary operations in LED are reduced without degrading the detection accuracy. The proposed method requires 25% to 35% less logic resources than the conventional method when implemented in an FPGA with a slight increase in latency.