The visualization of an image collection is the process of displaying acollection of images on a screen under some specific layout requirements. Thispaper focuses on an important problem that is not well addressed by theprevious methods: visualizing image collections into arbitrary layout shapeswhile arranging images according to user-defined semantic or visualcorrelations (e.g., color or object category). To this end, we first propose aproperty-based tree construction scheme to organize images of a collection intoa tree structure according to user-defined properties. In this way, images canbe adaptively placed with the desired semantic or visual correlations in thefinal visualization layout. Then, we design a two-step visualizationoptimization scheme to further optimize image layouts. As a result, multiplelayout effects including layout shape and image overlap ratio can beeffectively controlled to guarantee a satisfactory visualization. Finally, wealso propose a tree-transfer scheme such that visualization layouts can beadaptively changed when users select different "images of interest". Wedemonstrate the effectiveness of our proposed approach through the comparisonswith state-of-the-art visualization techniques.
展开▼