基于OTSU算法提取寒区河流流冰密度研究

Extracting River Ice Concentration in Cold Regions Based on the OTSU Algorithm

为预防冬春两季旱区河流凌汛灾害的产生,应用OTSU算法计算河冰分布密度,实时监测河流断面冰凌动态变化.通过索尼FDR-AX100E 4K数码摄像机对黑龙江干流漠河段进行流冰过程定点拍摄,提取目标断面影像信息.对比分析3种图像分割方法(边缘检测、阈值分割和区域分割)和4种图像阈值选取方法(人工选择法、直方图技术法、OTSU算法和迭代法)的优缺点,最终基于OTSU算法,利用Matlab仿真软件编程,对顶帽变换算法处理后的432张样本图像进行阈值分割,计算河冰像素占图像总像素比例,得到河冰分布密度,并绘制河冰分布密度随时间变化的动态曲线图.最终计算结果与现场实际观察冰凌变化情况吻合较好,表明应用OTSU算法计算河冰分布密度准确有效,实现了对目标断面的长时间连续监测.

To prevent the occurrence of ice flood disasters along a river in arid in winter and spring,the OTSU algorithm was used to calculate the ice concentration and monitor the dynamic change in ice in the river section in real time.The target section image information was extracted through fixed-point shooting of Mohe section of Heilongjiang river with a Sony FDR-AX100E 4K digital camera.The advantages and disadvantages of three image-segmentation methods(edge detection,threshold segmentation and region segmentation)and four image threshold-selection methods(artificial selection,histogram technology,the OTSU algorithm and the iteration method)were compared and analysed.Finally,based on the OTSU algorithm and MATLAB simulation software,432 sample images processed by the top hat transform algorithm were threshold segmented,the proportion of river ice pixels in the total image pixels was calculated,the river concentration was obtained,and the dynamic curve of river ice concentration with time was drawn.The final calculation results were in good agreement with the ice changes observed in the field,which shows that the OTSU algorithm is accurate and effective in calculating the ice concentration of the river,and the long-term continuous monitoring of the target section is realized.