| In recent years, the capability of computer hardware processing has gained a lot of development. However, traditional central processing video network is still inadequate to process the video network which is with huge amount of data. “Smart video processing node” is presented in this paper is a kind of smart platform which is running various computer vision algorithms for video image processing. It performs the actual algorithm processing tasks as the front-end video node. And then, the video image after processed will be sent to other nodes for future processing or sent to the back-end video server for display across the network. Numbers of smart video processing node consists of a kind of visual sensor network which is in distributed processing capabilities. And this video sensor network can effectively improves the performance and effectiveness of many traditional centralized video networks.This paper analyzes the structure of the VSNs and the roles which the smart video processing node plays in the VSNs. we propose the video processing related functional requirements based on the roles. In this paper, these video processing modules will be compiled into a separate dynamic linking libraries and static link libraries and we define these specific function libraries as the video processing middle wares for the smart video processing node. We use this middle wares to build the bridge combined the basis system and upper applications.FFMPEG audio and video processing open-source library is being used as the framework to achieve the video encoding and decoding goals; According to the requirements of distributed processing and collaborative computing in the video network, we achieve the target that divide the original video data into four horizontal and vertical video sub-blocks at any point in the video frame according the specific algorithm; We use SDL multimedia development library to display the decompressed video data; According PGM reliable multicast protocol, control instructions transmission is be achieved by one to many; Based on TCP protocol, task files and video streaming data are transfer by a kind of reliable way; Finally, we develop the wide field of two-way high-definition video camera panorama stitching software as the verification system. This software achieves the goal that video image seamless real-time stitches by calling the above-described dynamic linking library for video processing modules,which verifies the effectiveness and performance of the video processing middle wares.Making the analysis depended on the operating results of the verification system, the middle wares including control instructions transmission, compression, task file and video data transferring, decoding, video display runs well, what we developed achieves the desired goals. |
PDF Full Text Request