6533b85ffe1ef96bd12c2307

RESEARCH PRODUCT

Spatiocolorimetric neighborhood hypergraph and Image Processing Applications : Noise Removal and Edge Detection.

subject

description

In this document, we are interested in image modeling by the means of the hypergraph theory. Our contribution is essentially centered on the determination of the properties resulting from this theory and on the analysis from their adequacy with image problems, particularly edge and noise detection.First, we study the image spatiocolorimetric neighborhood hypergraph representation. Three representations are respectively presented incorporating global properties, local properties and similarity functions. Then, we use the hypergraph properties generated by the representation in order to define the structural models of noise and edge. This enables us to deduce the algorithms of noise suppression and edge detection on gray scale and color images. The performances of the proposed approachesare compared with the solutions classically used. Finally, the representation by neighborhood hypergraphconsistently seems to be efficient in low level image processing.