Method for automatic creation of a database of images searchable by its semantic content

摘要

Method in which a number of segmentations (131) are made from an initial image (4) into homogenous components, based on use of different criteria for identification and isolation of several object series constituted by the segmented images (41-43) associated with geometric descriptors of boxes surrounding the objects. The results of the segmentation process are fused together to form an indexed image database.