视频质量并行化分析系统

PARALLELISED ANALYSING SYSTEM FOR VIDEO QUALITY

视频质量代表了内容服务提供商的品牌形象,因此视频在上传至服务器前必须通过质量审核。海量的视频审核会带来巨大的人工成本,通过机器来辅助人工审核视频质量也成为了一种发展趋势。但随着视频解析度的不断上升,计算复杂度随之提高,传统单核计算的模式受到了很大的限制。提出一个通过并行化分析的框架,将视频质量的分析工作同时拆解到多个CPU上进行,并实现了相应的系统,大大提升了视频质量审查的速度。

Video quality represents the brand image of content services provider, therefore the video must be passed the quality auditing before to be uploaded onto the server. Massive video auditing will bring huge labour cost, which leads to the trend of manual video quality au- diting to be aided by the machine. However, as the video definition increases, the computing complexity is also raised. Traditional pattern of single-core computation is greatly limited. In this paper, we propose a framework for parallelised analysis, which dismantles the video quality analysing work onto multiple CPUs to process simultaneously. We also implement the corresponding system, it greatly improves the speed of video quality auditing.