Main Catalog Informatics, Computer Engineering and Control System Analysis, Control, and Information Processing

Image-matching algorithm using key points with scalability and rotation of objects

Authors: Suprun D.E.	Published: 12.10.2016
Published in issue: #5(110)/2016
DOI: 10.18698/0236-3933-2016-5-86-98
Category: Informatics, Computer Engineering and Control \| Chapter: System Analysis, Control, and Information Processing
Keywords: image matching, invariant characteristics, pyramid of Gaussians, Difference of Gaussians, key point, local extremum, feature vector, key point descriptor

This article describes an image-matching algorithm using key points to extract distinctive features invariant to scale and rotation. My work presents a feature extraction algorithm for reliable comparison of the different positions of the object or scene with a significant distortion range, changes in the 3D view, presence of noise and illumination changes. The scale-invariant feature transform (SIFT) method was used for matching images. The paper offers method implementation based on the pyramid of Gaussians and the Difference of Gaussian (DoG). The algorithm provides an oportunity to find the local extreme point, to detect key points, to build a feature vector and to compare local descriptors for further image pair matching under conditions of rotation, overlap, scale, change in point of shooting or lighting.

References

[1] Gaganov V. Invariant algorithms comparison point features in images. Komp’yuternaya grafika i mul’timedia [Computergraphics and Multimedia], 2009, no. 7 (1) Available at: http://cgm.computergraphics.ru/issues/issue17/invariant_features

[2] Grigor’ev Yu.A., Revunkov G.I. Banki dannykh [Databases]. Moscow, MGTU im. N.E. Baumana Publ., 2002. 318 p.

[3] Suprun D.E., Matveev V.A. The algorithm for creating a virtual mini-museum. Vestn. Mosk. Gos. Tekh. Univ. im. N.E. Baumana. Priborostr. [Herald of the Bauman Moscow State Tech. Univ., Instrum. Eng.], 2013, no. 4, pp. 67-78 (in Russ).

[4] Pope A., Lowe D. Probabilistic models of appearance for 3-D object recognition. International Journal of Computer Vision, 2000, vol. 40 (2), pp. 149-167.

[5] Meng Y. Implementing the Scale Invariant Feature Transform (SIFT) Method. Computer Science Department University of British Columbia Vancouver, B.C., Canada, 2006.

[6] Brown M., Winder S., Szeliski R. Multi-image matching using multi-scale oriented patches. Conference on Computer Vision and Pattern Recognition, 2005, pp. 510-517.

[7] Lowe D. Distinctive image features from scale-invariant key points. Computer Science Department University of British Columbia Vancouver, B.C., Canada, 2004.