气液两相流气泡图像的形态学分割方法

Morphological Segmentation Method for Bubbles Image in Gas-liquid Two-phase Flow

针对气液两相流图像中的气泡粘连现象,提出了一种用于识别粘连气泡的图像分割新算法。首先利用差影算法消除背景噪声,利用最佳阈值分割法将其二值化;然后采用形态学填充算法对气泡中心区域进行填充,并利用填充图像减去原二值化图像,从而分离出气泡的亮点图像;最后用形态学条件加厚算法对每个亮点进行重复加厚处理,得到气泡的分割图像。试验结果表明,该算法可以对气液两相流中的粘连气泡图像进行有效的分割识别,以便进一步准确地测量气泡尺寸及截面含气率。

Aiming at the phenomenon of bubble adhesion in the gas-liquid two-phase flow images ; a new algorithm of image segmentation to identify the bubble adhesion is presented. Firstly, the image-subtraction algorithm is used to eliminate the background noises and an optimal threshold segmentation algorithm is applied to obtain the binary images. Then, the central area of bubbles is filled with the morphology algorithm, and by subtracting the original binary images from filled images, the bright dots images of bubbles are extracted. Finally, the morphological conditions thickening algorithm is used to repeat thicken every bright dot to get segmental images. Tests results show that the algorithm can effectively segmental and identify the bubble adhesion images in gas-liquid two-phase flow, so that the size of bubble and gas rate in section can be measured aeeurately.