【作者】 刘荣丽；

【导师】 戴光清；

【作者基本信息】 四川大学 ， 水力学及河流动力学， 2003， 硕士

【摘要】 本文突破了点测量和机械探头传感器等传统的掺气浓度测量方法，采用非接触测量法，首次将计算机图像处理技术应用于水流掺气浓度的实测。基于水和气体对光的反射特性的不同，提出了一种自动识别水气二相流中的气泡的数字图像处理方法，并进一步给出计算二维、瞬时掺气浓度场的算法。流场试验结果分析表明：本文的方法完全可应用于水流掺气浓度场的实测，且效果较好。 文中分别采用激光片光源和影视灯照明，采用逐行扫描数字CCD摄像机获取气泡的流动图像；根据气泡和水对光的反射特性不同，提出采用阈值法识别图像中的气泡和水，从而为进一步计算掺气浓度打下坚实的基础。 阈值确定的准确与否直接影响掺气浓度的计算精度，在查阅大量文献和理论分析研究的基础上，并通过试验研究，本文提出了采用最大直方图熵法、类间方差法和迭代法等三种推求阈值的方法。经过大量的图像分割实验，对各种分割方法的分割效果和适用条件进行了比较和归纳，以计算精度和计算速度作为比较标准，作者认为类间方差法是计算阈值的最佳方法。文中还提出了一种改进的最大类间方差法，该算法核心思想就是：对于像素数为零的灰度级所对应的类间方差值指定为零，而不计算方差，从而大大减少了计算方差的次数，提高了原有算法的效率。 在上述工作的基础上，本文给出了计算二维、瞬时掺气浓度的算法，并编程予以实现。 四川大学工学硕士学位论文 结合面向对象的Visualc++高级编程技术，开发了灰度变换、图像增强、气泡识别、掺气浓度计算为主要功能的掺气浓度图像分析系统软件，显示了面向对象编程技术的应用优点。 为验证本文给出的掺气浓度图像处理方法的可靠性和实用性，作者采用图像处理技术对掺气浓度场进行实测研究，研究结果表明:文中实测计算出的掺气浓度值是可靠的，基本上能反映真实的掺气浓度情况，这表明本文提出的方法具有实际应用价值。 文中还对采用图像处理计算掺气浓度过程中出现的误差可能来源及提高测量精度的途径进行了分析和探讨，这对于提高掺气浓度的计算精度具有重要意义。 本文所作的工作为图像处理技术与水气两相流实测研究相结合的初步尝试和探索，研究内容体现了多学科前沿领域的交叉。研究成果为今后进一步深入研究和推广计算机图像处理技术在水气两相流测量中的应用，打下了有力的基础;对提高水流掺气浓度的测量水平和精度，进而提高水力模型试验水平、满足工程设计要求等都具有学术意义和重要实用价值。更多还原

【Abstract】 Based on image process, a new method is proposed to measure void fraction without contact. This method breaks through the traditional point measure and sensor technology.According to different optics-reverberation character of water and gas, this paper put forward how to identify gas from the gas-water digital image and how to f -urther calculate void fraction. The result shows that the method possesses relatively satisfactory precision.At first, bubble flow image is acquired using laser and incandescence lamp respectively as light source and using CCD. According to different optics-dispersion character, threshold segmentation method is proposed to identify gas from the gas-water flow, which make steadfast base for further calculate void fraction.The precision of threshold directly affect the calculate precision of void fraction. Via refer to literature and theory study, three methods, namely ENT, OTSU, and iteration method are putted forward to calculate void fraction and via lots of experiment, three methods are compared and induced. In the end, OTSU is recommend which it has satisfactory precision and speed.Based on the former result, the algorithm of two dimension and instantaneous void fraction is proposed and carry out through programming.United with Visual C++ language which face to object, a software is exploitedwhich have functions like gray transform, image improve, bubble identify, voidfraction calculation, which show the strongpoint of object-facing programming.In order to validate the reliability and practicability of the image technique in measurement of bubble void fraction, a test aiming at water-gas field is done. The result shows that the value of void fraction is reliable and can reflect the real status.In this paper the error source is analyzed and discussed and the approach of how to advance measure precision is proposed, which has significant fact meaning.The content in this paper is preparatory attempt that combines image technique and gas water field measurement technology. This research make strong base for the further application of image technique in two phases flow of gas-water and has significant learning and practicable value.更多还原