Gaussian noise elimination in colour images by vector-connected filters

Abstract

This paper deals with the use of vector-connected filters for eliminating Gaussian noise in colour images. This class of morphological filters suppresses noise but preserves the contours of the objects. We impose a total order between pixels for morphological processing. Once the HSI space has been adapted, we employ it in the lexicographical order. As such, all of the morphological operations are vectorial. After having defined the vectorial geodesic operators, they are then employed to eliminate Gaussian noise.