MEMORI

The application aims at detecting objects in visual material in order to provide a richer content-based indexing scheme, which exploits the semantic information contained in the images. Object detection is achieved by segmenting the input image and grouping the obtained regions by means of a process driven by the recognition. Given a database of objects, each described by a 3D model or a complete set of views of the object itself, the system builds an indexed database of objects views which acts as an objects recognizer. The groups hypothesis are validated by the object classifier and only the groups that more likely correspond to an object of the database will survive. The result is a set of object detection hypothesis along with a correspondent set of confidence indices. This result will be used to provide a richer description of the input image that will be exploited in a content-based image retrieval system. That main current limitations are given by the restriction on the number and nature of objects recognized by the system, and the processing time