基于图像处理的三峡库区水域水华检测算法及预警研究

Research on water bloom detecting algorithm and early warning based on image processing for Three Gorges Reservoir area

针对传统水华监测方法存在的过分依赖人工操作、实时性差、成本高、普及性差、主观性大、劳动强度大等缺陷,提出了一种可以自动检测三峡库区水域水华现象的图像处理算法,并将该算法应用于库区水域水华预警。即利用历史数据建立颜色先验模型,运用改进的Camshift算法检测水华区域,通过区域面积的变化分析水华灾情趋势,并建立水华预警模型进而对水华预警。实验结果表明该算法应用于水面视频监控可以自动检测出水华区域并有效地对水华实时预警。

Traditional water bloom detecting methods have the features such as over-reliance on manual operation,low real-time,high cost,poor popularity,large subjective,labor-intensive and other defects.Aiming at these defects,an algorithm based on image processing technology was proposed,which can automatically detect water blooms in Three Gorges Reservoir Area,and the algorithm was applied to water bloom early warning in the reservoir area.The algorithm uses historical data to establish a color prior model,uses modified Camshift algorithm to detect the region of water blooms,obtains disaster trends according to the change of the area of water blooms,and builds a water bloom early warning model;at last,the water bloom warning is given.Experimental results show that when applied to water surface video surveillance the algorithm can automatically detect the region of water blooms and can realize real time water bloom early warning.