Publication Type : Journal Article
Publisher : Communications in Computer and Information Science
Source : Communications in Computer and Information Science, Volume 283 CCIS, Number 1, Gandhigram, Tamil Nadu, p.413-420 (2012)
ISBN : 9783642289255
Keywords : Feature extraction, Feature points, Mathematical models, Panorama, Principal component analysis, Saddle, SaddleSURF, Scale spaces, Statistical tests, SURF
Campus : Coimbatore
School : School of Engineering
Department : Computer Science
Year : 2012
Abstract : This paper presents a modified Speeded Up Robust Features (SURF) with feature point detector based on scale space saddle points. Most of the feature detectors like Scale Invariant Feature Transform (SIFT), Principal Component Analysis (PCA)-SIFT and SURF are based on extrema points i.e. local maxima and minima. This work aims at utilizing the saddle points for panorama stitching which is a common and direct application for feature matching. Here Euclidean distance of descriptor is used to find the correct matches. Experiments to test the performance are done on Oxford affine covariant dataset and compared the performance with that of SURF. © 2012 Springer-Verlag.
Cite this Research Publication : S. S. Kecheril, Issac, A., and C. Velayutham, S., “SaddleSURF: A saddle based interest point detector”, Communications in Computer and Information Science, vol. 283 CCIS, pp. 413-420, 2012.