Abstract—Although a face image can be described in
illumination-reflectance model, the illumination and reflectance
components are not distinctly separated in the frequency
domain after the Fourier transformation. In fact, the
illumination components of a poorly illuminated face image can
be divided into two regions, namely low frequency region and
vertical center region. This paper proposed two methods to
identify the locations of illumination component in the face
images which consist of different subjects and illumination
directions. By using inspection method and mean square error
(MSE) function, the illumination locations can be determined
accurately. The largest location of illumination component
discovered via the research will be used as reference parameters
to design a customized filter before it is implemented in
homomorphic filtering technique.
Index Terms—Face images, frequency domain, fourier
spectrum, homomorphic filtering technique, illumination,
reflectance.
Face images, frequency domain, fourier
spectrum, homomorphic filtering technique, illumination,
reflectance.
[PDF]
Cite:Pin Yang Tan and Haidi Ibrahim, "A Research on the Location for Illumination Component of
Poorly Illuminated Face Images in Fourier Spectrum," International Journal of Modeling and Optimization vol. 3, no. 6, pp. 495-498, 2013.