Comparison of optical flow algorithms performance on flame image sequences

Analyzing flame image sequences of co-combusting pulverized coal and biomass can be used to determine the state of this very complex process. Considering dynamic changes between successive frames of such video streams the optical flow algorithms could be applied for diagnostic purposes. Effectiveness and performance of two approaches (sparse and dense optical flow) are discussed, applied for several video streams of co-combusting biomass and pulverized coal, recorded with different combustion settings with a high-speed camera at the laboratory stand.