Subject Content-Based Intelligent Cropping of Digital Photos

Image cropping is one of the most important operations performed to enhance photographs. The direct benefit is improved image composition in terms of better subject location, better subject magnification, and reduced background clutter. Indirectly, cropping can also help image and subject rendering because unwanted distractions are eliminated before image enhancement. We present a robust and efficient solution to this challenging problem for consumer snapshot photos. First, a main subject detection algorithm is employed to produce a belief map probabilistically indicating the subject content. Facilitated by the use of an integral image, a globally optimum cropping window is located to maximize the subject content within the cropping window while satisfying multiple constraints. Extensive evaluation using multiple judges has shown its overwhelming advantage over a commercially used cropping scheme.