In a cavitating water jet, cavity clouds emerge and collapse with an unsteady, but periodic tendency where the frequencies depend on the working conditions. The presented work aims at examining and analyze the dynamic behavior and properties of the clouds under different circumstances. Computer vision and image processing were introduced as tools to define the cavitation clouds based on the Contour Recognition technique. A Canny operator and Otsu threshold fragmenting methods were used. The use of these methods allows for a better understanding of the cavitating jet clouds' behavior based on the pixel intensities and shows that for an arbitrary cloud the surface itself has a dynamic feature and depends on the cavity composition. The clouds' properties could be measured and correlated to the applied working conditions. Also, the oscillation frequencies of the area of the clouds could be determined. The analysis shows that the quality of the obtained results depends mainly on the input threshold values separating the foreground and background pixels. The difficulty of defining the threshold value is discussed in the paper, as well as the validity of using the Contour Recognition technique in this field.

在空化水射流中，空腔云以不稳定但周期性的趋势出现和坍塌，其中频率取决于工作条件。所提出的工作旨在检查和分析不同情况下云的动态行为和特性。计算机视觉和图像处理被引入作为基于轮廓识别技术定义空化云的工具。使用了 Canny 算子和 Otsu 阈值分割方法。使用这些方法可以更好地理解基于像素强度的空化喷射云的行为，并表明对于任意云，表面本身具有动态特征并取决于空腔组成。可以测量云的特性并将其与应用的工作条件相关联。还，可以确定云区域的振荡频率。分析表明，所得结果的质量主要取决于分隔前景和背景像素的输入阈值。论文讨论了定义阈值的难度，以及在该领域使用轮廓识别技术的有效性。