Search

Assuming the dichromatic image model we propose a global reduction of specularity effects by means of parametric illumination gradient images obtained by fitting 2D Legendre polynomials to the specular component of the images. Fitting is done applying a (λ + μ) Evolution Strategy. The method could be applied to static robotic monitoring in teams of robots, where the illumination gradient image could be computed once and applied to successive frames until the illumination conditions change drastically. The method could be useful for the detection of image regions with different chromatic properties. [ABSTRACT FROM AUTHOR]