End-to-End Perceptual Quality-Of-Experience Measurement and Optimization to Reshape Modern Video Communication Systems

随着“体验时代”的到来，丰富和增强我们在电视、平板电脑和智能手机上消费数字视频内容的感知体验已成为现代视频分发系统的核心目标。然而，最近的调查表明，终端用户糟糕的视频体验质量（QoE）已经成为视频服务生态系统的主要威胁。这提出了一个紧迫而普遍的研究挑战，视频发布者需要有效的机制，不仅可以帮助获得个人消费者的即时质量体验，还可以执行战略优化，以经济有效的方式提高视觉质量体验。随着超高清、高动态范围、宽色域和高帧率视频以及为增强现实和美德现实观看体验而创建的视频等新型媒体格式的日益普及，这一挑战变得更加明显。

With the arrival of the “era of experience”, enriching and enhancing perceptual experience in consuming digital video content delivered to our TVs, tablets and smartphones has become a central goal of modern video distribution systems. However, recent surveys indicated that the poor video quality-of-experience (QoE) of end users has become a major threat to the video service ecosystem. This presents an urgent and pervasive research challenge, where video distributors require effective mechanisms that can not only help gain instantaneous awareness of the QoE of individual consumers, but also perform strategic optimizations to improve visual QoE in a cost-effective way. The challenge becomes even more pronounced with the growing popularity of novel media format such as ultra high-definition, high dynamic range, wide color gamut, and high frame rate videos, as well as videos created for augmented reality and virtue reality viewing experiences. The long-term research objective of this project is to develop theories and critical technologies for unified end-to-end visual QoE measurement, optimization and intelligence solutions that, if successful, will reshape how videos are distributed through global multimedia distribution networks and meanwhile improve everyone's everyday viewing experience. Specifically, we will develop a series of objective QoE measurement algorithms that produce consistent QoE measures in real-time for various media content types. We will use these novel QoE measures to optimize video distribution system designs, focusing on QoE-driven video coding, streaming and quality enhancement. We will also develop deep learning technologies for QoE assessment and QoE-driven optimizations. Subjective QoE testing will be conducted to validate and improve the performance of the algorithms being developed. The project will be carried out by a world leading research team that has a strong track record of making outstanding research contributions, evidenced by numerous prestigious awards and recognitions as well as publications in top journals and conferences with top citations. The team also has extensive experience collaborating with leading industrial partners and organizations. This project will help strengthen Canada's reputation, impact and global leadership in the field of image processing and multimedia communications. It will also help produce a significant number of HQPs who are expected to join Canada's workforce in academic, industrial and government R&D organizations. The developed technologies will be transferred to Canadian industrial partners' products and services. As such, Canadian companies will benefit significantly for increased revenue and improved competitiveness in the global market, and meanwhile create new jobs in the information technology sector. Thus it has the potential to make a lasting impact on Canada's economy.