Normalization of 2D Images “in Terms of Level” Based on Cosine and Hadamard Transform
DOI:
https://doi.org/10.20535/1810-0546.2016.1.51768Keywords:
normalization in terms of level, image classification, Hadamard transform, cosine transform, pattern recognitionAbstract
Background. The problem of pattern recognition is considered in this paper. We propose to use orthogonal basis functions for pattern recognition of 2D signals.
Objective. The aim of this paper is investigation of pattern recognition problem of 2D images based on cosine and Hadamard transforms.
Methods. Pre-processing of signals is proposed to reduce the number of spectral components in the orthogonal series of signal decomposition. Normalization of a reference signal “in terms of level” allows comparing reference and tested signals by means of transform coefficient calculation. This coefficient is a match criterion of these signals.
Results. Theoretical information used for proposed classification method is described. The normalization algorithms of the reference and tested signals, which have to perform before determining similarity of images, are proposed. Transformation of a reference signal to two-dimensional basic function is fundamental point. This basic function is based on selected 2D orthogonal transform. Complete cycle of calculation for pattern recognition of some 2D images is executed. The calculations were performed using Hadamard and cosine transforms. 2D images of obtained spectra are given.
Conclusions. As a result, this research shows that we can perform pattern recognition of normalized 2D images based on calculation of transform coefficient. The analysis of calculation results shows that classification of signals based on transform coefficient values is possible. Development of matching classifiers is the aim of the next research.
References
V.G. Abakumov and A.I. Rybin, Biomedical Signals (Genesis, Treatment, Monitoring).Kyiv,Ukraine: Nora-Print, 2001, 516 p. (in Ukrainian).
A.N. Prodeus and Ye.N. Zakhrabova, Expert Systems in Medicine.Kyiv,Ukraine: VEK+, 1998, 320 p. (in Russian).
W. Ouyang et al., “Fast pattern matching using orthogonal Haar transform”, in Proc. of IEEE Conf. on Computer Vision and Pattern Recognition, San Francisco, CA, USA, June 13–18, 2010, pp. 3050–3057. doi: 10.1109/CVPR.2010.5540058
A.I. Rybin et al., “Transform coefficients of normalized orthogonal transformations and diagnosis of pulsegrams”, Visnyk NTUU KPI. Ser. Radiotehnika. Radioaparatobuduvannya, no. 30, pp. 148–156, 2005.
A.I. Rybin and A.D. Melnyk, “Matched normalized signal filtering”, Izv. Vysh. Uchebn. Zaved. Radioelektron., vol. 51, no. 2, pp. 112–114, 2008. doi: 10.3103/S0735272708020106
A.I. Rybin, “Normalization of discrete orthogonal transformations by a test signal”, Izv. Vysh. Uchebn. Zaved. Radioelektron., vol. 47, no. 7, pp. 27–31, 2004.
A.I. Rybin and Yu.Kh. Nizhebetskaya, “Analysis of images similarity and difference using normal orthogonal conversion”, Izv. Vysh. Uchebn. Zaved. Radioelektron., vol. 53, no. 3, pp. 167–172, 2010. doi: 10.3103/S0735272710030076
Downloads
Published
Issue
Section
License
Copyright (c) 2017 NTUU KPI Authors who publish with this journal agree to the following terms:- Authors retain copyright and grant the journal right of first publication with the work simultaneously licensed under CC BY 4.0 that allows others to share the work with an acknowledgement of the work's authorship and initial publication in this journal.
- Authors are able to enter into separate, additional contractual arrangements for the non-exclusive distribution of the journal's published version of the work (e.g., post it to an institutional repository or publish it in a book), with an acknowledgement of its initial publication in this journal.
- Authors are permitted and encouraged to post their work online (e.g., in institutional repositories or on their website) prior to and during the submission process, as it can lead to productive exchanges, as well as earlier and greater citation of published work
