Navigation and Browsing of Image Databases

摘要

Image collections are growing at an exponential rate and solutions to manage vast databases of images are hence highly sought after. Content-based image retrieval techniques have shown great potential, yet commonly employed approaches like query-by-example are only of limited usefulness. An interesting alternative is provided by systems that allow visual exploration of an image dataset through a browsing interface. In these methods the complete database, or parts thereof, is visualised through application of dimensionality reduction techniques, clustered visualisations or display of a graph structure. Once visualised, it should then be possible to browse through the collection in an interactive, intuitive and efficient way. In this paper we present various browsing techniques that can be employed for this purpose. Browsing can be achieved in several ways. We can distinguish between horizontal browsing which works on images of the visualisation plane, and includes operations such as panning, zooming, magnification and scaling, and vertical browsing which allows navigation to a different level of a hierarchically organised visualisation. Furthermore, browsing can also be accomplished by taking into account time stamp information, hence enabling temporal browsing. We conclude, highlighting the need for objective evaluation and benchmarking of browsing system and see one of the next research challenges in the development of effective image browsing tools for mobile devices.