Structured layout for resizable background art

We have investigated a method to represent dynamically resizable background art images using a structured layout language. Background art is often used in photo layout to enhance the photo sharing experience, to impose a theme in story telling or to add better visual effect to photos. Different from natural scene photographs, background art images tend to have strong regularity and symmetry. Directly applying general resizing techniques to the whole image often fails to preserve the original design intention. By examining the background images that are manually composed for different page sizes by graphic artists, we have come up with a set of design theories. Based on these theories, we developed a design language and a transformation algorithm that together enable dynamic adaptation of images to different layout dimensions, while preserving the original look and feel. The design language describes the composition of a background with primitive image elements. Based on the attributes of the elements, the transformation algorithm dictates how each element should be resized and translated with the layout dimension. In our test cases, the automatically resized images give comparable results to manually resized ones. This method provides several benefits for automatic photo album layout: not only it adapts the art design to different page dimensions, it also automatically adjusts the photo placement regions, and it often allows the composition of large and high resolution images with relatively small number of image primitives.