An Optimal Fusion Marker-Based Watershed Segmentation Method For Bubble Metering

In the industrial flotation process, visual characteristics of the foam, such as its size and quantity, are key indicators for assessing and adjusting the flotation state. Thus, accurate segmentation of foam images to extract this statistical information is particularly crucial. However, effective segmentation of foam images remains a challenge due to the variability in foam characteristics caused by different raw ore properties and operational conditions. Despite numerous image segmentation methods proposed by researchers, difficulties such as high-lit areas in the images, adhesion between foams, and image noise are still challenging to overcome. In view of this, this paper presents a flotation foam image segmentation method based on optimal fusion marking. This method combines a block sub-graph classification model with boundary constraints and introduces optimal fusion marking to achieve more precise segmentation of foam images. Finally, the effectiveness and superiority of this method are verified through a series of experiments.