生物式水质监测系统若干关键技术的分析与实现-analysis and implementation of some key technologies in biological water quality monitoring system.docxVIP

杭州电子科技大学硕士学位论文 杭州电子科技大学硕士学位论文 I I 摘 要 当前的常规理化式水质监测不能对水中含有的未知有毒物质进行监测，也不 能对几种污染物的联合毒性进行监测.生物式水质监测方法克服了理化监测的局 限性，具有良好的实际应用前景。本文在前人研究的基础上，应用计算机视觉和 视频跟踪技术，通过对斑马鱼活动轨迹的跟踪和各项运动指标的分析，实现水质 监测的目的。 本论文主要进行了以下几个方面的研究工作： 1.研究了以斑马鱼为受试生物的生物式水质监测系统的系统设计。提出了生 物式水质监测系统的需求分析，研究了水质监测系统的整体设计目标和整体功能 区划，研究了水质监测系统在机械层、电子层和软件层等层面的设计与实现。 2.研究了水质监测系统中用到的视频跟踪和滤波技术。分析了几种主流的视 频跟踪方法，提出了适于水下鱼类跟踪的目标跟踪方法。论文还研究了应用卡尔 曼滤波对跟踪的搜索范围进行缩小的方法，并结合水下鱼类跟踪的具体特点，对 卡尔曼滤波器的参数进行了重新设置，提高了跟踪算法的效率。 3. 研究了以行为指标变化为核心的水质综合预警技术，根据污染物投放和 鱼行为变化的对应关系，提出了平均速度、平均高度、平均距离、活鱼条数“四 位一体”的水质监测综合预警体系，并把亨克利统计算法应用于鱼类水质监测领 域。最后本文通过毒性试验，初步证明了该预警体系的有效性、稳定性和鲁棒性， 为下一步优化和改良生物式水质监测系统打下基础。 关键词：生物式水质监测，视频跟踪，卡尔曼滤波，水质预警 II II ABSTRACT Traditional technologies based on physics and chemistry can’t discover unknown toxic substance in water, they can’t test the whole toxicity when several pollutants are mixed together. Biological water quality monitoring methods counteract these weaknesses and have a good future. This paper designed a new kind of biological water quality monitoring system, based on zebra fish’s fast reaction to the pollutions. We have done the research in several areas: Research on how to design and build a biological water quality monitoring system which used zebra fish as its testing animals. We do a demand analysis of this kind of system, Then we discussed the design goals and the functional areas of our system. We researched on the mechanical layer, the electronic layer, and the software layer of the system in detail. Research on the video tracking technology used in the monitoring system. We researched the difficulties of video tracking, several mainstream methods of video tracking, their features and their limitations. We researched the filtering method of video tracking ——Kalman filtering, which is put into practical monitoring with the feature-based object tracking method. Propose the water quality pre-alarming method. Four indictors of zebra fish movement were designed, including average speed, average height, average distance