Suchergebnisse

9 pages, 9 figures.-- ©2003 Society of Photo-Optical Instrumentation Engineers. ; We present a new approach to achieve object identification based on the use of phase correlation in the scale transform domain for automatic character recognition. The results are extensible to other fields. The proposed method is shown to be invariant to translation, rotation, and scale. We extended the methodology used by Casasent and Psaltis by considering a more efficient digital scale transform as an alternative to the Fourier-Mellin techniques. To improve the discriminative power, we introduce a new template matching based on the use of a modified weighted log-polar spectrum. The correlations have been calculated by using phase-only filters (POF) in a digital system. The proposed method is able to provide discrimination between scale and rotation in images to facilitate image registration. ; This work was partly supported by the European Union under the MAS3-CT97-0122 ADIAC project (Automatic Diatom Identification and Classification) and INCO-DC961646 AMOVIP project (Advanced Modeling of Visual Information Processing). The authors acknowledge the financial help of a bilateral co-operative programme: CSIC(Spain)-CONACYT(México). Part of this work was supported by Conacyt with the project "Procesado automático de partículas biogénicas" with reference number: 36075-B. ; Peer reviewed