Compression of multiple user photo galleries

The paper presents a predictive coding strategy for multi-user photo gallery, which initially localizes each picture in terms of viewpoint, orientation, time, and acquired elements. This information permits ordering all the images in a prediction tree and associates to each of them a reference picture. From this structure, it is possible to build a predictive coding strategy that exploits the redundant elements between the image to be coded and its reference. Experimental results show an average bit rate reduction up to 75% with respect to HEVC Intra low complexity coding.