On the morphological processing of hue

Authors:
E. Aptoula;S. Lefèvre
Affiliations:
LSIIT UMR-7005 CNRS-ULP, Pôle API, Blvd Sébastien Brant, P.O. Box 10413, 67412 Illkirch Cedex, France;LSIIT UMR-7005 CNRS-ULP, Pôle API, Blvd Sébastien Brant, P.O. Box 10413, 67412 Illkirch Cedex, France
Venue:
Image and Vision Computing
Year:
2009

Citing 5
Cited 1

Why mathematical morphology needs complete lattices

Signal Processing
Morphological operators for image sequences

Computer Vision and Image Understanding
Digital Image Processing

Digital Image Processing
A comparative study on multivariate mathematical morphology

Pattern Recognition
Morphological operators on the unit circle

IEEE Transactions on Image Processing

Granule-view based feature extraction and classification approach to color image segmentation in a manifold space

Neurocomputing

Quantified Score

Hi-index	0.00

Visualization

Abstract

Although polar colour spaces are being increasingly used in the context of colour mathematical morphology, mainly due to their intuitiveness, the processing of the circular hue band continues to be their main drawback. In this paper, we discuss the two principal problems concerning the morphological processing of hue, first its lack of a lattice structure, which we propose to introduce by means of a distance based formulation from multiple representative reference hue points. And second, the distinction of chromatic from achromatic pixels, since the hue component is of no significance for ''low'' saturation levels. For this purpose, we present a new weighting scheme, based on the combination of saturation and luminance channels. Application results on texture classification, asserting the superior performance of this approach, are also included.