PROJECT TITLE :

Manifold Learning and Spectral Clustering for Image Phylogeny Forests

ABSTRACT:

The ever-increasing variety of gadgets being used to create digital content, still as the easiness in sharing, editing, and republishing this content, brings the problem of coping with a massive amount of digital objects (e.g., images or videos) whose content is very similar. Some problems faced by investigators of digital crimes when analyzing this type of information embody finding the initial source of a suspect image, and the accountable for first publishing it. It is conjointly challenging to work out how these objects are related to each other. Recent efforts in developing algorithms to find automatically the underlying relationship among teams of digital media objects with similar content are explored within the multimedia phylogeny field. A tree structure is employed to represent the link among these objects, galvanized by the phylogenetic trees in biology. Discovering whether these objects came from the identical source or from different sources is basically a clustering downside: one) connected objects belong to the identical cluster (tree) and a pair of) unrelated objects should match in several clusters. During this paper, we have a tendency to address the problem of finding these clusters in sets of semantically similar pictures, prior to tree reconstruction. We have a tendency to propose the mix of manifold learning and spectral clustering approaches, which are successfully employed in totally different applications embedding the first knowledge into a lower, but meaningful, dimensional house. Experiments with a lot of than 40 00zero take a look at cases show that the proposed approach improves the accuracy in finding the proper variety of trees within the set, also because the reconstruction of the phylogeny trees.