Gaussian noise removal by color morphology and polar color models

Abstract

This paper deals with the use of morphological filters by reconstruction of the mathematical morphology for Gaussian noise removal in color images. These new vector connected have the property of suppressing details preserving the contours of the objects. For the extension of the mathematical morphology to color images we chose a new polar color space, the l1-norme. This color model guarantees the formation of the complete lattice necessary in mathematical morphology avoiding the drawbacks of others polar spaces. Finally, after having defined the vectorial geodesic operators, the opening and closing by reconstruction are then employed for the Gaussian noise elimination.