Adapting Surneillance Video to Small Displays Via Object-Based Cropping

The increasing availability of small mobile devices capable of playing video, such as PDAs and phones, makes possible their use for new video applications. In these small resolution devices, adaptation is essential to maximize the utility of the delivered content improving the user experience. Content analysis usually helps to detect the interesting parts trying to focus the adaptation on them. In this paper, we propose adaptive cropping as a useful method to deliver adapted video with the objects of interest for the user. Window filtering is used to cope with detection errors, and also to remove annoying artefacts as flickering and unnatural motion in the adaptation window. Two cases are studied: fixed size adaptation window and fixed aspect ratio adaptation window. Although the algorithms presented are generic, in this paper we focus on video surveillance sequences.