A study of high-qualityInternet broadcasting using video contents

这项研究的目标是使用视频内容提供高质量的互联网广播。为此，我提出了一种参考视频内容的流媒体服务的码率控制方法，其控制目标是空间分辨率、时间分辨率和帧大小。在这项研究中，我将视频节目归类到大多数用户感觉相同视频质量的组中，并分析了影响主观视频质量的因素(质量参数)。此外，将分析结果应用到实际的编码系统中，并演示了该方法在实际流媒体服务器上的实现过程。在本研究中，我获得了以下三个结果。网络播出视频节目分类：利用30个不同类型的视频节目，对28名被试进行了主观评价，提出了将视频节目分为6组的方法。该分类方法用户特征描述参数是语义参数(例如，语音的重要性)和视觉参数(例如，运动强度)的组合。设计传输速率控制算法：通过多元线性回归分析，将各组主观质量值与质量参数值之间的关系表示为方程，并利用该方程设计了质量控制算法。通过选择在约束条件下最大化主观质量值的参数值，可以实现有效的质量控制。将该方法应用到一个实际的编码系统中：使用TMPGEnc(即一个MPEG-1编码器)作为一个实际的编码系统，得到了所提出的控制算法的约束条件。在这种情况下，我采用编码比特率作为代价函数(约束)，并提出了一种在约束下最大化主观质量值的控制方法。

The goal of this study is to provide high-quality Internet broadcasting using video content. For this purpose, I proposed a rate control method for streaming services with reference to video content, which control targets are spatial resolution, time resolution and frame size. In this study, I classified video programs into some groups in which large majority of users feel the same video quality, and analyzed factors (quality parameters) affecting subjective video quality. In addition, I applied the analysis results to an actual encoding system and demonstrated the implementation process of the proposed method on practical streaming servers. In this study, I obtained the following three results.1. Video program classification for Internet broadcasting :I conducted subjective assessments for 28 subjects using various types of 30 video programs, and proposed a method of classifying video programs into 6 groups. The classification method users feature description parameters which are the combination of semantic parameters (e.g. the importance of voice) and vision parameters (e.g. degree of motion intensity).2. Designing a control algorithm of transmission rate :I expressed the relationship between the subjective quality value of each group and the quality parameter values as an equation by multiple linear regression analysis, and designed a quality control algorithm using the equations. Efficient quality control can be achieved by selecting parameter values that maximize the subjective quality value under constraints.3. Applying the method to an actual encoding system :I used TMPGEnc (that is an MPEG-1 encoder) as an actual encoding system, and obtained the constraint for the proposed control algorithm. In this case, I adopted encoding bit-rate as the cost function (the constraint), and proposed a control method maximizing the subjective quality value under the constraint.