基于视觉分析的淤积河道流量测量方法的研究

【作者】 王平；

【导师】 常发亮；

【作者基本信息】 山东大学 ， 控制理论与控制工程， 2006， 硕士

【摘要】 传统的河道侧流方法大部分是依靠测流船、铅鱼等设备通过人工定时读取流道进、出口处水尺得到水位值，然后通过查有关河道形状、流态等参数对应的曲线取修正参数后，借助于经验公式，通过计算得到流量值。这种方法不仅费时费力，不能连续或实时计量，而且其准确性也受到人为因素等影响。尤其是对于像黄河这种多泥沙河流，由于受到淤积情况的影响，其引水渠道河底形成淤积，渠道底面形成不规则的形状，往往得不到较为理想的流量曲线，进而影响测量结果，用传统的测流方法无法完成全自动测流。本文针对这一情况，提出了一种计算机自动视觉和测流铅鱼相结合的流量测量的新方法，解决了淤积河道无法全自动测流的问题。利用计算机视觉在子像素分辨率下自动分析河水表面的宽度，控制铅鱼沿横向进行多点的水深和流速的测量，从而得到淤积河道河水横断面的形状，依此计算河水的瞬时流量。该方法已经在黄河引水河道中得到了成功的应用，应用结果表明了该方法的有效性。 本文以位山引黄涵闸为实例，介绍基于视觉分析的淤积河道流量测量新方法。系统地介绍了本课题的的设计思想、硬件构成和软件实现，其中重点叙述对视觉图像的分析和处理。首先根据河边图像的特点，采用迭代阈值法对图像进行二值化，对二值化后的图像利用形态学方法进行滤波并进行细化处理，除去非边缘图像，得到动态水面的边缘图像，再采用HOUGH变换求得子像素分辨率下的河面边缘和标志杆的直线方程，从而自动分析河水表面的宽度。再根据河面宽度，确定测点数及测点位置，利用PLC控制测流铅鱼，测量水深和瞬时流速，从而计算出瞬时流量。系统根据控制要求还可以实现三种不同的工作方法，即远程自动测流和远程半自动测流及现场手动测流三种方式，三种测流方式的测流数据(瞬时流量、累计流量、流速等)都能够自动进入计算机系统。 系统由PLC控制测流铅鱼进行实时测量，并将数据传送到上位机进行处理，本文介绍了PLC输入输出信号的分配、采样信号的获取以及PLC与上位机的通讯，给出了较为详细的接线图、程序流程图。更多还原

【Abstract】 Mostly we read the water gauge at inlet and outlet of the flow path depending on devices such as Measurement Ship and Lead Fish, and then calculate the flux value recurring to experiential formula by consulting curves with parameters relating to figure of flow path and flow patterns. This is the traditional method of channel side flow measurement. Such method is not only time consuming and power consuming, incapable of continuous or real time computation but also easily influenced in veracity by contrived factors. Especially for Yellow River whose channel is silted up with mud and sand, sometimes we cannot obtain perfect curves of flux value and influence the results sequentially. This is for reason that influence of filling up situation results in filling up of diversion water channel, then anomalistic shapes are formed on channel underside. So the traditional flow measurement method is incapable of automatic flow measuring. Taking into consideration such situation, a new flow measurement method is presented to solve the automatic flow measurement problem of silted channel by using computer vision and Lead Fish. The river surface width is automatically analyzed under the sub-pixel resolution by using the computer vision, and the river depth and flow velocity of multi-points are measured by controlling the Lead Fish to move along the landscape orientation. So the cross shape of the silted river can be obtained, and the on-time flow can be calculated accordingly. This method has been successfully applied in diversion channel for Yellow River, and the practical measuring results have proved that the method is effective and reliable.The WeiShan Diversion Culvert-Brake is presented as an example in the article and a new flow measurement method based on machine vision analysis used for silted channel is introduced. The design idea of the task, the structure of hardware and the realization of software are introduced systematically. The analysis and processing for vision image is given focus depiction. Firstly the iterative threshold method is adopted to binary the image according to the characteristic of river water. The filtering algorithm and thinning algorithm are applied in binary images by using morphology method. After that the non-edge images are removed and the edge images of dynamic water surface are obtained, then the linear equation of river surface edge and sign perch is calculated under sub-pixel resolution by using HOUGH algorithm. So the width of river surface can be analyzed automatically. Next we fixed on the number and position of measurement points, control the flow measurement Lead Fish by PLC and calculate instantaneous flow by measuring water depth and instantaneous velocity of flow. Three distinct更多还原