Mobile multimedia: From acquisition to adaptation with semantics, context and social information
MetadataShow full item record
The booming development of mobile devices and wireless networks has brought about a new era of ubiquitous mobile multimedia. The pervasiveness and explosion of mobile media content invokes an on-going revolution that shifts PC centralized media capturing, sharing and consuming to the mobile user centralized mode for nearly every type of applications. However, the rise of mobile media applications also imposes new challenges, from the way media is captured and transmitted, to the way people interact with the mobile device, and to the way the media is consumed by people. Direct application of PC based approaches does not produce the desired results. The unique characteristics of mobile media have to be treated differently. Meanwhile, the additional information available in mobile media applications enables new possibilities to provide an enhanced user experience. In this dissertation, we focus on improving mobile users' experiences in several new media applications. The research works presented in this dissertation encompass several stages in mobile multimedia, from mobile media acquisition to media adaptation and consumption stages on mobile devices. At the media acquisition stage, it is a challenging task for common mobile users to capture high quality photos without professional photography knowledge. To address the pervasive challenge in mobile photography, we first design a socialized mobile photography scheme to assist mobile users in capturing high quality photos by crowdsourcing learning from social media. The proposed approach is able to mine composition knowledge and exposure principles to guide photographing on mobile devices. To assess the quality of photos acquired by mobile devices, we also develop in this dissertation a novel scheme in photo quality assessment, a key component in evaluating media acquisition for various mobile media research works. This new scheme is capable of being aware of the context information associated with mobile media. We achieve this by leveraging the context and visual content to learn relevant aesthetic rules for an improved quality assessment. At the media consumption stage, we take the constraints in mobile media transmission and presentation into full consideration. We design an elegant adaptation scheme for viewing consumer photos on remote mobile devices. This scheme delivers to the mobile users their desired media content based on semantics and user preferences to improve the user's media viewing experience. The next research work focuses on the compact presentation of social media feeds, which dominates contemporary mobile media consumption. To address the challenges in viewing explosive social media feeds under mobile display constraints, we develop an innovative system to automatically generate social media snippets for efficient mobile browsing. Based on aesthetic study and visual perception principles, key information is extracted and optimally composed to generate the social media summary in an appealing, attractive and informative fashion to help mobile users grasp contents of interest quickly and easily from the vast sea of social media.