Quality-Aware Predictor-Based Optimal Adaptation of Still-Image Multipart Multimedia Messages

Multimedia Messaging Services (MMS) allow users with heterogeneous terminals to exchange structured messages composed of text, images, sound, and video. The MMS market grows rapidly, posing the problem of MMS adaptation, necessary to ensure interoperability between terminals. Message adaptation poses technological challenges, especially when one considers the case of high-volume service providing. In this work, we propose novel predictor-based dynamic programming approaches to MMS adaptation, explicitly maximizing user experience, rather than relying on heuristics to deliver satisfactorily adapted messages. We show that the proposed solutions lead to noticeably superior image quality with faster transcoding times than comparative algorithms inspired by what is found in actual products and in the literature.