Images play an important role in sharing, expressing and exchanging information in our daily lives. Today, the typical user takes hundreds of photos that are then stacked in a hard disk. However, too often the photos are not labeled properly and remain unorganized, which results in enormous collections of photos that are difficult to skim through.
The focus of this project was to develop new ways of photo searching in a personal collection and give users a better experience when browsing them. In order to simplify image navigation, an unsupervised image clustering framework was implemented through an adaptation of the k-means algorithm. The clustering method is based on hierarchical image grouping using content-based and metadata features. Also, a new visualization and interaction method of image browsing is proposed for mobile devices. Experimental results demonstrated the good performance of the proposed methods on a real image database.
Author: Pedro Teixeira
Type: MSc thesis
Work done at: Faculdade de Engenharia da Universidade do Porto