Image retrieval using linear greyscale granulometries

Opening distributions for image analysis constitute an extremely useful tool in morphological tasks. Efficient techniques have been proposed to compute granulometries in greyscale images based on linear openings. Greyscale granulometries using opening distributions are equivalent to a contour extraction problem. The present study addresses the development of a morphological approach based on particle size distributions for the analysis of greyscale images using linear openings and their use as an input to build queries on a image database. From the experimental results, we show that the pattern spectra derived from the images can be used in texture based image retrieval.