无线多媒体传感器网络中基于多类传感器信息的分布式视频处理

Multi-modal Sensors Information Based Distributed Video Processing for Wireless Multimedia Sensor Networks

无线多媒体传感器网络中的多媒体传感器节点的功率、计算和存储能力及带宽资源受限,迫切需要低复杂度的视频处理技术对大数据量的视频进行压缩传输。基于来自网络中监控同一目标场景的多种类型传感器的信息,给出了一个新的无线多媒体传感器网络分布式视频处理系统框架;提出了一种基于多类型传感数据融合和多视角的GOP(group of picture)划分方法;在解码端,考虑单一视角视频序列之间较强的时间相关性,产生时间相关的边信息;利用来自多个视频传感节点的视频序列间的空间多视角相关性,产生多视角相关的边信息,并提供两种边信息的融合和选择机制,提高边信息的准确度和可靠性。最后仿真实验结果表明该方法的有效性和优越性。

In wireless muhimedia sensor networks （WMSN） , sensor devices are constrained in terms of-power, processing, memory and bandwidth capability. Low-complexity video processing technique is highly desired to encode and transmit large numbers of video data. Based on the information from multi-modal sensors which monitor the same scene, a new framework of distributed video processing system for WMSN is presented. A group of pictures （GOP） partition algorithm based on multi-modal data and multi-view is also put forward. In the decoder, considering the temporal correlation between adjacent frames from single view, temporal side information is generated. Multi-view side information is acquired by using spatial correlation among the frames from adjacent video sensors. In order to improve the prediction accuracy and reliability, a mechanism to fusion and select two kinds of side information is provided. The validity and advantage of the proposed method are verified through simulations.