Dynamic prediction of interface level using spatial temporal markov random field

Abstract Maintaining a desired interface level plays a key role for the efficient extraction in oil sands, mining and related industries. However, varying throughputs and downstream disturbances tend to change the interface level over time. Hence, measurement of the interface level is an important indicator of the process behavior and is useful for economical operation and improved control. In this paper, we propose a Spatial Temporal Markov Random Field (ST-MRF) based image segmentation approach for the dynamic prediction of the interface level. We assume that interface level, which is modeled as MRF, propagates as a Markov chain across time, yielding a ST-MRF model. The proposed approach is validated using the images captured from laboratory scale equipment designed to simulate the industrial scenarios.