基于分水岭变换的浮选泡沫图像分割

An Image Segmentation of Flotation Froth Based on Watershed Transformation

针对煤泥浮选泡沫图像中煤泥气泡互相粘连、边界模糊的情况,根据分水岭变换模拟浸水原理,提出一种新的、有效的分割算法,同时应用数学形态学中的腐蚀和膨胀算法,解决了过分割和欠分割问题,使得粘连的煤泥气泡得到了有效的分割。并对分割后的图像进行了物理特征提取,如气泡的数目、颜色、大小分布等,这些特征参数是评价浮选过程好坏的重要信息,利用它们可监测和优化浮选过程。该分割算法成功的用于煤泥浮选泡沫的分割,并且提取出了煤泥泡沫图像所有静态的物理特征参数。

A new and efficient image segmentation algorithms is developed for the coal flotation froth images-the bubbles are connected and have the murky verge between them, based on watershed transformation. At the same time, the added algorithm based on mathematical morphology has solved the problems of over-segmentation and under-segmentation of the connected components. The size, color, shape of froth bubbles are the very important information for production optimization in the coal flotation. In order to determine these parameters, bubbles in the froth images should be delineated. This algorithm successfully segments the coal flotation bubbles and gets all the physical parameter of the off-line froth flotation of coal.