Research And Application Of Distributed Video Processing Platform Based On Stream Computing

As a key application of the internet of things, the real time video processing platform provides users a variety of services based on video content. Video processing platform has its own features:big data, complex data structure, high transmission load. As a result, traditional method with single node, multiple threads can hardly meet the requirements. So, it’s important in researching distributed video processing platform based on stream computing.Stream computing is the key technology of cloud computing, application modules can make use of computing resource as demand; memory based database and message queue can make full use of high speed memory, provide data transmission and interactive services in high speed. On the basis of integrate analysis of stream computing, memory based database and message queue, we designed distributed video processing platform based on stream computing. Firstly, a video data acquisition module based on ZeroMQ was presented to collect large amounts of video data. Secondly, video processing applications was realized on the Storm platform, which takes advantage of Storm platform in real time and parallelized data processing. Besides, we realized application management structure to operate different video applications on the same platform. Finally, video acquisition and image matching application was deployed. The result of video acquisition and processing shows distributed video processing platform is suitable in real time and huge video data processing.