建立多样化、智能化的传感网水环境监测体系是当前的发展趋势。已有的传感网监测系统侧重于对水质指标参数的检测，依据阈值或统计模型进行预警评估，缺少对水面污染直观有效的监测手段。本项目针对水环境监测中的水面污染物探测与污染源定位追踪问题，开展传感网视觉分析与移动节点协同监测技术研究。建立统一的水面污染物检测识别框架，能有效克服水面波纹、倒影、颜色区分度差等因素的影响，自动实现对动态水域漂浮物的检测与覆盖以及对静态水域污染物区域的识别，提取出水面图像中漂浮物与污染区域。综合考虑传感网链路质量与节点剩余能量，研究水面图像感兴趣区域压缩传输算法。联合网络连通性、信息收益与节点能耗约束，研究多节点协同监测模型与运动控制策略，实现对水面污染源的连续定位追踪，并预测污染源的动态扩散过程。项目研究成果能满足水环境监测多样化与分析智能化的需求，有助于构建高效、合理的水环境监测预警体系。

It is a developing trend to establish more diversified and intelligent water environment monitoring systems based on sensor network technology. The existing sensor networks-based water quality monitoring systems focus on the measurement of water parameters, and carry out early warning and evaluation according to the alarm threshold or statistical models, which lacks direct and effective monitoring means for surface pollution. As for water environment monitoring based on sensor networks, this project places emphasis on visual analysis and mobile node collaborative monitoring for the detection of surface pollutants and pollution source tracing. A unified framework for the detection and identification of surface pollutants that can effectively overcome the influence of water ripples, reflections and color difference will be established to automatically implement the detection and coverage of floating objects in dynamic waters and the identification of pollutant regions in static waters. Considering the link quality and residual energy of nodes, an image compressive transmission method for the extracted areas of interest will be investigated. Combining with network connectivity, information benefit and energy consumption constraints, the multi-node collaborative model and motion control strategy will be further studied to realize continuous positioning and tracking of the pollution source as well as predict the dynamic diffusion process. Research results of the project can meet the needs of monitoring diversification and analysis intelligence, and contribute to the construction of efficient and reasonable water environment monitoring and early warning systems.