A New Invariant to Illumination Feature Descriptor for Pattern Recognition
- 106 Downloads
Abstract—A new descriptor for describing features in gray-scale images that is invariant to nonuniform illumination is proposed. The suggested method for the feature descriptor design is based on a local energy model which is a biologically plausible model of the visual system. The algorithm for feature detection and construction of the descriptor uses the scale-space monogenic signal framework and a modified algorithm for calculation of the histogram of oriented gradients based on the phase congruence of the signals. The results of computer simulation show that the proposed descriptor provides excellent detection and matching of features at nonuniform illumination, noise, and minor geometric distortions in comparison with known descriptors.
This work was supported by the Russian Science Foundation, grant no. 15-19-10010.
- 5.D. H. Hubel, J. Wensveen, and B. Wick, Eye, Brain, and Vision (Sci. Am. Library, New York, 1995).Google Scholar
- 10.Morrone M. Concetta and D. C. Burr, “Feature detection in human vision: A phase-dependent energy model,” Proc. R. Soc. B: Biolog. Sci. 235 (1280), 221–245 (1988).Google Scholar
- 12.P. Kovesi, “Image features from phase congruency,” J. Comp. Vision Res. 1 (3), 1–26 (1999).Google Scholar
- 13.P. Kovesi et al., “Edges are not just steps,” in Proc. 5th Asian Conf. Comput. Vision, (ACCV 2002), Melbourne, 2002, Vol. 8, pp. 22–8.Google Scholar
- 17.C. Harris and M. Stephens, “A combined corner and edge detector,” in Proc. Alvey Vision Conf., CiteSeer, 15, 10–5244 (1988).Google Scholar
- 19.J. Diaz-Escobar and V. Kober, “A robust hog-based descriptor for pattern recognition,” SPIE Opt. Eng.+ Appl., pp. 99712.Google Scholar
- 22.H. Bay, T. Tuytelaars, and L. Van Gool, “Surf: Speeded up robust features,” in Proc. Eur. Conf.on Computer Vision, 2006 (Springer-Verlag, 2006), pp. 404–417.Google Scholar