Image Phylogeny Forest Construction using Manifold Learning and Spectral Clustering

A simple search for an image on the Web can return thousands of related images. Some of them are original copies, others may be the variants of the same digital image, and others are unrelated. It is straightforward to detect exact image duplicates; this is not the case for slightly modified versions. Some issues faced by investigators of digital crimes when analyzing this type of data include finding the original source of a suspect image, and the one responsible for first publishing it. It is difficult to determine how these objects are related to each other. Recent efforts to find automatically the underlying relationship among groups of digital media objects with similar content have been explored in the multimedia phylogeny field. The relationship among these images is represented using tree structure. Discovering whether these images came from the same source or from different sources is a challenging problem. The proposed method addresses the problem of finding these clusters in sets of semantically similar images, prior to tree reconstruction. The combination of manifold learning and spectral clustering approaches, which have been successfully used in different applications embedding the original data into a lower, but meaningful, dimensional space.