New Approach for Fast and Accurate Color‐pattern Recognition
- Conference date: 21–26 October 2007
- Location: Campinas, São Paulo (Brazil)
Measurement of the reflectance spectrum at each spatial point of an object surface certainly provides high accuracy for color classification and recognition but it takes time and requires huge computer resource. The method of principal components analysis allows us to find an optimal (low‐dimensional) set of basic spectra with which any reflectance spectrum can be reconstructed with any required precision. By using these basic spectra high compression of multispectral images can be achieved. In this report we describe novel system for optical implementation of the principal component analysis. Projections of a reflection spectrum on the preliminary defined set of basic spectra (spectral eigenvectors) are directly and simultaneously measured at all spatial points of the object surface. The system is capable to implement even spectral eigenvectors with negative components. Key‐element of the proposed approach is the computer‐controlled light source. It consists of a set of the light‐emitting diodes and generates any predefined basic spectrum with possibility of fast switching from one spectrum to another. The system can measure two‐dimensional distribution of reflection spectra in whole visible diapason. Thereafter, the measured data are used for fast and accurate color‐pattern recognition without reconstruction of the original reflection spectrum. Feasibility of the proposed technique is experimentally demonstrated by measuring weighting coefficients for four metameric colors.
- Visible spectra
- Computer hardware
- Optical devices
- Spatial analysis
- Spectrum analysis
MOST READ THIS MONTH
MOST CITED THIS MONTH
Y. K. Semertzidis, M. Aoki, M. Auzinsh, V. Balakin, A. Bazhan, G. W. Bennett, R. M. Carey, P. Cushman, P. T. Debevec, A. Dudnikov, F. J. M. Farley, D. W. Hertzog, M. Iwasaki, K. Jungmann, D. Kawall, B. Khazin, I. B. Khriplovich, B. Kirk, Y. Kuno, D. M. Lazarus, L. B. Leipuner, V. Logashenko, K. R. Lynch, W. J. Marciano, R. McNabb, W. Meng, J. P. Miller, W. M. Morse, C. J. G. Onderwater, Y. F. Orlov, C. S. Ozben, R. Prigl, S. Rescia, B. L. Roberts, N. Shafer‐Ray, A. Silenko, E. J. Stephenson, K. Yoshimura and EDM Collaboration
Article metrics loading...