全景视频QoE评价研究进展

随着5G移动通信技术、高性能计算、传感技术的不断进步，全景视频受到了越来越多的关注。全景视频通过头戴显示设备，可以为用户提供远超于平面视频的逼真的立体视觉感知，具有良好的发展前景。为用户提供良好的体验质量（Quality of Experience，QoE）是视频服务提供商吸引和留住用户，在激烈的市场竞争中取得成功的关键。与平面视频相比，全景视频的数据量倍增，对视频数据的采集、编码、传输和存储均提出了更高的要求。因此，如何在网络传输带宽、存储资源有限的情况下，保证用户的QoE就成为工业界和学术界共同关注的研究热点问题。本文对全景视频QoE评价进行了综述，首先对全景视频在采集、拼接、投影、编码、传输、解码、反投影、渲染等各个环节可能存在的失真进行分析，总结归纳了用户QoE的各种影响因素，如人的因素、系统因素、情境上下文和视频内容特性等；在此基础上，从影响因素和建模方法等多个方面归纳了全景视频QoE评价模型的研究进展，及其在码率自适应、资源优化分配和码率控制等方面的应用情况；最后，介绍了具有代表性的全景视频QoE评价数据集以及常见的QoE模型性能评价准则，并探讨了QoE评价模型目前存在的问题和未来的研究方向。 

Survey on QoE Evaluation of Panoramic Video

? ?With the continuous advancement of the 5th Generation （5G） mobile communication technology， high-performance computing and sensing technologies， panoramic video has attracted more and more research attention. Through head-mounted display devices， panoramic video can provide users with realistic and immersive visual perception far exceeding that of flat video， and has a promising prospect. Providing users with a good Quality of Experience （QoE） is the key for video service providers to attract and retain users and succeed in the fierce market competition. Compared with the usual flat video， the data volume of panoramic video usually increases exponentially， which will put forward higher requirements for the acquisition， coding， transmission and storage of video data. So how to ensure users’ QoE with limited network transmission bandwidth and storage resources has become a hot research topic in both academia and industry recently. In this paper， the QoE evaluation of panoramic video is mainly reviewed. Firstly， the possible distortion types of panoramic video in the process of video acquisition， stitching， projection， encoding， transmission， decoding， back projection， rendering， and so on， are analyzed comprehensively， and a variety of subjective and objective influencing factors of the users’ QoE are summarized， such as human factors， system factors， context， and video content characteristics and so on. On this basis， the research advances of the QoE evaluation model of panoramic video are highlighted from the aspects of influencing factors， modeling methods and so on. And the main applications of the QoE evaluation model of panoramic video in the Adaptive Bit Rate （ABR） algorithm， resource optimization allocation， bit rate control， etc， are analyzed. Finally， several representative QoE evaluation datasets of panoramic video and commonly used performance evaluation metrics are introduced. The existing problems and the future tendency of the QoE evaluation model of panoramic video are also further discussed.