Title :
Improved Harris algorithm within scale-invariance
Author :
Yongtao, Hao ; Ping, Yu
Author_Institution :
CAD Res. Center, Tongji Univ., Shanghai, China
fDate :
Nov. 30 2010-Dec. 2 2010
Abstract :
Based on the SIFT recognition method, this solution is to obtain the natural features in computer vision recognition and matching by means of the Gauss pyramid, DOG pyramid, their differential functions appoximated instead of σ2∇2G to achieve the scale invariance practice. Through the Harris corner detection algorithm for the corresponding transformation formula, and the Gauss pyramids, DOG pyramid, the modified algorithm obtain a certain degree of scale invariance. This improved algorithm can be used either to supplement characteristics of the SIFT algorithm, and can also be used as a separately kind of scale invariance of the feature recognition algorithm.
Keywords :
Gaussian processes; differential equations; edge detection; feature extraction; DOG pyramid; Gauss pyramid; Harris corner detection algorithm; SIFT recognition method; computer vision recognition; differential functions; feature recognition; DOG pyramid; Feature recognition; Gaussian pyramid; Harris corner detection; scale invariance; sift;
Conference_Titel :
Computer Sciences and Convergence Information Technology (ICCIT), 2010 5th International Conference on
Conference_Location :
Seoul
Print_ISBN :
978-1-4244-8567-3
Electronic_ISBN :
978-89-88678-30-5
DOI :
10.1109/ICCIT.2010.5711139
