气液两相辨识中一种类圆形目标的分类方法

Elliptical objects classification method in gas-liquid recognition

化工设备中气液两相辨识过程中的气液接触面积是一个需要获得的关键参数。目前用图像处理和分析的方法获取这一参数成为研究的一个热点。本文将气泡图像的识别归结为类圆形目标的分类。由于分类的依据是气泡表面积的大小 ,确定了圆面积和圆形度两个最能表征类圆形表面积的分量组成分类特征向量。然后用模糊聚类的方法实现目标的分类 ,并根据实际数据的特点灵活地选择和调整了模糊聚类的距离函数 。

Gas\|liquid contiguity area is a key parameter in the course of gas\|liquid recognition in chemical equipment. To obtain this parameter by image processing is becoming the focus of research. In this paper, the recognition of bubble image is summed up as the classification of elliptical objects. Area and circularity act as two elements of the eigenvector for they can remarkably characterize the surface area of the bubbles. Objects are classified by using fussy clustering method. Distance function is selected and adjusted in conformity to the character of practical data. Elliptical objects are classified concisely.