Photo Layout with a Fast Evaluation Method and Genetic Algorithm

Abstract

Photo collages are an effective form of visual communication and sharing. This paper is devoted to a particular style of photo layout, in which photos are tightly packed into a rectangular canvas while the aspect ratio and orientation of photos are preserved. In addition, users may specify a desired size for each photo within the canvas. BRIC (block recursive image composition) is a method that has been proposed for this problem [1]. In this paper, we present a better method based on the framework of genetic algorithm. Our method builds on three key components. First, we developed a fast method for computing the photo dimensions for a given layout tree. This makes it feasible to evaluate a large number of layout trees within a time limit that is acceptable for interactive multimedia applications. Second, we adapted a fitness measure that takes into account both the photo coverage of the canvas area and the size distribution of photos. Finally, we constructed appropriate genetic operators and incorporated the aforementioned evaluation method and fitness measure. Our experimental results showed that the proposed method is consistently better than BRIC.