Kalman Filtering Applied to Sheet Measurement

Sheet-making processes such as aluminum foil, plastic sheet, or paper generally employ on-line sensors to measure the thickness or weight per area of the sheet. The sensors may be placed at a fixed distance from the edge, scanned over the sheet, or may provide non-scanning full width measurements. The character of the process variations vary widely according to the nature of the process. Some contain strong periodic components, while others are more random. The goal of signal processing for sheet measurement is to produce a precise estimate of sheet variations. In this paper, we describe how the Kalman filter can be applied to estimate the variations in different types of sheets. For metal sheet, the periodic variations caused by rolls are our main concern. For plastic sheet, narrow streaks on the sheet need to be detected immediately. Two dimensional estimation is crucial to paper sheet measurement. A dual Kalman filter which contains the combination of a temporal model and a spatial model is proposed to produce the required estimate.