Browsing Large Image Datasets through Voronoi Diagrams

Abstract

Fig. 1. An example of a thumbnail bar, shaped as a belt that partially encircles the title of this paper. It features irregular shaped thumbnails with gradually decreasing sizes, starting from the currently selected image (in the middle). Shown thumbnails represent a dynamic subset of a large dataset composed of a few thousands of images; the selected subset is adaptive, denser around the selected image and respectful of existing image hierarchies. When a new image is selected, the thumbnails dynamically rearrange around the new selection in a configuration preserving all the above characteristics and without breaking temporal continuity.Abstract-Conventional browsing of image collections use mechanisms such as thumbnails arranged on a regular grid or on a line, often mounted over a scrollable panel. However, this approach does not scale well with the size of the datasets (number of images). In this paper, we propose a new thumbnail-based interface to browse large collections of images. Our approach is based on weighted centroidal anisotropic Voronoi diagrams. A dynamically changing subset of images is represented by thumbnails and shown on the screen. Thumbnails are shaped like general polygons, to better cover screen space, while still reflecting the original aspect ratios or orientation of the represented images. During the browsing process, thumbnails are dynamically rearranged, reshaped and rescaled. The objective is to devote more screen space (more numerous and larger thumbnails) to the parts of the dataset closer to the current region of interest, and progressively lesser away from it, while still making the dataset visible as a whole. During the entire process, temporal coherence is always maintained. GPU implementation easily guarantees the frame rates needed for fully smooth interactivity.

Towards interactive, intelligent, and integrated multimedia analytics

“…Quadrianto et al introduce a multiple-tier semantic hierarchy [68], with the grid in each successive tier showing representative images for an increasingly detailed neighbourhood of interest. The approach by Brivio et al utilizes Voronoi diagrams to fill the display, with the focus item placed in the center and other items being represented on the screen based on their distance to the focus with respect to an ordering of the collection [7]. Overall, similarity-based approaches combine screen space efficiency with an enhanced notion of structure through visually contiguous regions.…”

iGraph: a graph-based technique for visual analytics of image and text collections

“…Examples include PhotoMesa with quantum treemaps and bubblemaps for zoomable image browsing, 3 PhotoTOC for automatic clustering for personal photograph browsing, 4 spiral and concentric rings for focus+context visualization, 5 MoireGraphs with the radial graph layout for visual exploration, 6 semantic image browser (SIB) with the multidimensional scaling layout for image overview, 7 and dynamic image browsing with Voronoi diagrams. 8 Companies like Google and Microsoft also developed products such as Image Swirl and Bing Visual Search for organizing large image collections. Besides images, text visualization has received lots of attention recently.…”

A Compositional Model for Low-Dimensional Image Set Representation

“…It can also provide an efficient way to navigate large datasets by automatically ordering the set w.r.t. a specific variation parameter [4]. Finally, this may have potential applications in detection and recognition problems [1].…”