6533b827fe1ef96bd1287570

RESEARCH PRODUCT

An Integrated Method for Image Retrieval

Di Gesù V.Domenico Tegolo

subject

pictorial darabase feature extraction image segmentation information fusion.Settore INF/01 - Informatica

description

This paper presents an information fusion method for image retrieval; the retrieval strategy is based on statistical and geometrical features extracted from sub-images. Our goal is to find a set of best similar images related to a prototype image; this goal may be obtained according to image content rather than symbolic attributes. MRI (Magnetic Resonance Imaging) images and astronomical images have been adopted to test the method. Main steps of the procedure to retrieve images are: (l) Segmentation, (2) Matching, and (3) Decision. The first step involves four clusters co-operating segmentation algorithms; in the matching step, a sets of candidate similar images is provided; finally an information fusion decision rule is applied to extract the best matching to a given prototype image.

yearjournalcountryeditionlanguage
1995-01-01
http://hdl.handle.net/10447/229298