In the hot rolling process of aluminum, a roll coating is formed on the surface of the work roll. The roll coating consists of particles of aluminum, aluminum oxide, and coolant. When the roll coating is thick or nonuniformly distributed, it detaches from the work roll onto the surface of the rolled material, causing surface defects on aluminum products. Therefore, the condition of the roll coating must be properly managed. However, work roll monitoring techniques applicable in the hot and humid environment of the manufacturing process do not yet exist. To develop such monitoring techniques, we focus on the optical changes of the surface of the work roll and process these optical changes quantitatively by extracting image features from work roll images taken by a camera. The features we designed in this work were found to be effective in capturing the trend of optical changes of the work roll surface.
