Burst bubble recognition based on depth feature

The bursting rate of bubbles is an indicator of bubble stability. To accurately recognize burst bubbles, the Kinect sensor is first introduced to the froth flotation industrial site, then the bubble depth information is collected accurately. Due to the bubble burst is close related with depth change, depth feature is extracted by our proposed method based on depth difference. The proposed method filters potential burst bubbles and registers single bubble region based on color data, and then recognizes bubble bursts in terms of depth data. the experimental results show that our proposed method can effectively identify bubble bursts, the precision and recall reach 0.8547 and 0.8963 respectively.