Design And Implementation Of Video Real-time Transcoding System Based On Cloud Computing

With the rapid development of mobile Internet technology and the increasing popularity of mobile devices such as mobile phones and tablets,China's video service industry is developing rapidly.Because the user terminals have different network bandwidth,and the support formats for video playback are different,video service provider will transcode video,including conversion of encoding format,resolution,number of frames,etc.Video transcoding is divided into offline transcoding and real-time transcoding,offline transcoding is the transcoding of offline video files,and real-time transcoding is the transcoding of real-time video data streams,such as surveillance video and live broadcast video.Video transcoding system deals with source video,providing users with diverse video services.The traditional single point centralized video transcoding,using multitask processing method,is unable to meet the current massive video transcoding requirements.Therefore,the research of distributed video real-time transcoding system,based on the stream and real-time computing framework,is very necessary.This thesis studies parallel transcoding for massive multi-channel real-time video,by using the real time computing platform Storm in cloud computing and its stream processing mode.Storm as one of Apache's top open source projects,is used to compute and process real-time data,and it reduces latency in data processing.This video real-time transcoding system is divided into three parts: system management,Storm job cluster and video streaming service.In the system management section,his thesis combines the open source frameworks Jersey,Spring and Mybatis,and implement the background APIs of RESTful style,which are used for user management,video streaming management,transcoding template configuration and task management.In the storm cluster,the submitted running video-transcoding job deals with the video stream such as acquisition,fragmentation,transcoding and stream-pushing.Using storm programming model,distributed parallel transcoding is realized,and transcoding efficiency is improved.In the Streaming media service section,mainly for RTMP video transmission protocol,the thesis designs and implements a streaming media server,and develops a video player based on RTMP protocol.Finally,based on the video real-time transcoding system,the thesis completes the functional and performance tests.The thesis performs a test on two improved algorithms,then tests the effect of the transcoding concurrent numbers on the transcoding efficiency.