河域水陆分界线实时视觉识别方法研究

Research on Real-Time Visual Recognition Method of River Basin Boundary

为实现全自动化无人测流的目标,现通过高分影像摄像机实时获取现场数据,针对河域边界实时视觉识别进行了方法研究和实验分析。研究过程中,首先对研究对象进行预处理以得到初步降噪的灰度图,然后采用阈值分割法进行二值化处理以得到水陆二值图,在二值图的基础上进行种子填充来避免非水体中连通域噪声的影响及数学形态学运算使水陆分界线的边缘特征连续且清晰,最后采用canny算子进行边缘检测提取并细化处理,最终实现了对河域水陆分界线的实时视觉识别。实验结果表明,该方法可以有效地识别定位在碎石淤泥河沙等物复杂干扰下的水陆分界线,避免了目前测流工作中的人工目视识别定位水边线的弊端,为实现全自动化测流奠定坚实基础,可极大减少在河域测流工作中人力及财力的投入。

In order to achieve the goal of fully automated unmanned flow measurement,real-time field data is now acquired by high-resolution video cameras.Based on the MATLAB platform,it conducts method research and experimental analysis for real-time visual recognition of river boundaries.In the research process,the research object is first preprocessed to obtain a grayscale image with preliminary noise reduction,then threshold segmentation is used for binarization,and then seed filling and math⁃ematical morphology operations are performed on the basis of the binary image.The edge features are continuous and obvious.Fi⁃nally,the canny operator is used to extract and refine the edge detection,and finally realize the real-time visual recognition of the water and land boundary of the river.The experimental results show that this method can effectively identify the boundary line of water and land under the complex interference of gravel,silt,river,sand,etc.,avoiding the disadvantages of manual visual rec⁃ognition and positioning of waterline in current flow measurement work,and is in order to realize fully automated measurement.Laying a solid foundation for flow can greatly reduce the investment of manpower and financial resources in the work of measur⁃ing flow in the river basin.