中文摘要：

基于气泡形状特征是相关工业过程状况的重要标识,提出基于峰谷分析的气泡形状特征计算方法,通过对泡沫图像进行多维峰谷分析,统计在不同方向上气泡的总宽度,并构造与实际气泡同轴相似的超级气泡,然后用最小二乘法进行曲线拟合,以此为基础估算出气泡的平均偏转角度、平均离心率、平均尺寸、数量等形状特征。在仿真实验中,对泡沫图像进行0°,45°,90°和135°方向的均匀扫描及峰谷分析,并估计气泡的平均偏转角度、平均离心率、数量。研究结果表明：与人工的平均偏转角度、平均离心率、数量测量值相比,采用所提出的计算方法所得结果相对误差分别为4.99%,3.46%和31.1%;与其他气泡形状特征计算方法相比,本文所提出的方法具有结果准确、速度快、受光照环境和噪音影响较小的特点,适用于工业现场实时监控。

英文摘要：

Considering the importance of bubble shape features to identify the relevant industrial process conditions, a features recognition method for bubble shaping based on peak analysis was proposed. Firstly, bubbles widths computed by multidimensional peak analysis in different directions of foam image were summed to construct a super bubble with elliptical curve coaxial similar to real bubbles. Then the formula of super ellipse curve was fitted by the least square method to estimate the average deflection angle, eccentricity, size and number of bubbles. In the simulation, the uniform scanning and peak analysis of foam image were made in direction of 0°, 45°, 90° and 135°. The results show that compared to the artificial methods, the deviation of estimated average deflection angle, eccentricity, number of bubbles form manual measurement value obtained by the features recognition method are 4.99%, 3.46%, 31.1%, respectively. Compared to other feature recognition methods for bubble shape, the proposed method is accurate, fast and nearly unaffected by light and noise, and suitable for real-time monitoring in industrial site.