| Abstrakt: |
Vessels for the liquid metal have an inner lining of refractory bricks to withstand the high temperatures of molten steel. To counteract the wear of the lining, gunning is used as a common procedure in steelmaking to prolong the usability of the lining. During gunning, a concretelike, moist refractory mass is applied to the lining surface by pumping it through a nozzle and pointing the spouting mass jet at the desired area. The optimal machine settings, e.g., pressure levels and mixture ratio, are crucial to the maintenance quality. These parameters are reflected in the gunning jet appearance and can be observed and interpreted by experienced operators. Video tracking of the gunning jet facilitates remote observation and is key to automated feature analysis. In this article, we focus on tracking the jet using infrared images. In particular, we use a particle filter with online learning to update the reference template. Template matching is used for tracker stabilization and object verification. We use a parametric model of the gunning jet for feature extraction and dynamic reference template generation. For tracker initialization, we combine template matching with geometric- and temperature-based constraints. We provide experiments on real and synthetic data to demonstrate the capabilities of the presented tracker. [ABSTRACT FROM AUTHOR] |
