Froth image segmentation algorithm

Abstract

This paper describes a froth image segmentation algorithm. In order to determine the bubble size, shape and texture information from the froth images, the images have to be segmented. Because of the large variation of the froth image patterns and quality, it is difficult to use the existing segmentation algorithm. Therefore, we modified and combined different existing image segmentation algorithms to form a new segmentation algorithm for froth images. In the algorithm, we used and modified a thresholding algorithm to automatically detect the maximum bubbles, then perform a morphological technique to growing bubble regions. The over-segmentation problem was resolved by an object merging algorithm based on bubble shape analysis. In the paper, three froth image segmentation results are demonstrated, which show that the developed segmentation algorithm is reasonable for froth images.