Wireless line sensor network for distributed visual surveillance

Wireless sensor networks (WSNs) play a crucial role in visual surveillance for automatic object detection, such as real-time traffic monitoring, vehicle parking control, intrusion detection,and so on. These online surveillance applications require efficient computation and distribution of complex image data over the wireless camera network with high reliability and detection rate in real time. Traditionally, such applications make use of camera modules capturing a flow of two dimensional images through time. The resulting huge amount of image data impose severe requirements on the resource constrained WSN nodes which need to store, process and deliver the image data or results within a certain deadline. In this paper we present a WSN framework based on line sensor architecture capable of capturing a continuous stream of temporal one dimensional image (line image). The associated one dimensional image processing algorithms are able to achieve significantly faster processing results with much less storage and bandwidth requirement while conserving the node energy. Moreover, the different operating modes offered by the proposed WSN framework provide the end user with different tradeoff in terms of node computation versus communication bandwidth efficiency. Our framework is illustrated through a testbed using IEEE 802.15.4 communication stack and a real-time operating system along with one dimensional image processing. The proposed line sensor based WSN architecture can also be a desirable solution to broader multimedia based WSN systems.

无线传感器网络（WSNs）在自动目标检测的视觉监控中起着至关重要的作用，例如实时交通监测、车辆停车控制、入侵检测等。这些在线监控应用需要在无线摄像网络上高效地计算和分发复杂的图像数据，并具有高可靠性和实时检测率。传统上，此类应用利用摄像模块随时间捕捉二维图像流。由此产生的大量图像数据对资源受限的无线传感器网络节点提出了严格要求，这些节点需要在一定期限内存储、处理和传递图像数据或结果。在本文中，我们提出了一种基于线传感器架构的无线传感器网络框架，该框架能够捕捉连续的时间一维图像（线图像）流。相关的一维图像处理算法能够以更少的存储和带宽需求实现显著更快的处理结果，同时节省节点能量。此外，所提出的无线传感器网络框架提供的不同操作模式在节点计算与通信带宽效率方面为终端用户提供了不同的权衡。我们通过一个使用IEEE 802.15.4通信栈和实时操作系统以及一维图像处理的测试平台对我们的框架进行了说明。所提出的基于线传感器的无线传感器网络架构对于更广泛的基于多媒体的无线传感器网络系统也可能是一个理想的解决方案。