Abstract
This paper presents a detailed description of a vision system that detects and localizes the nonuniformities that appear on the paper slurry (wood fiber and water mixture) at the wet end of a paper machine. Specifically, the system monitors the paper slurry as it exits the headbox and alerts the operators of any event (e.g., streaks) that disrupts the otherwise homogeneous background. These events affect crucial product properties such as formation. A poor formation results in thick and thin spots on the sheet and impacts its strength and printability. This paper describes the vision system in terms of its hardware modules, as well as the image processing algorithms that it utilizes to perform its function. The system acquires intensity and topographic information from the scene. It uses textureβbased features for the detection and facetβbased descriptors for the localization of the nonuniformities. In addition to being tested in a laboratory environment, a prototype of this system was constructed and deployed to a paper mill, where its performance was evaluated under realistic conditions. Installed on a fourdrinier paper machine, running at 480 m/min and producing linerboard material, the vision system monitored about a 1 m wide area and successfully detected and localized slurry streaks. Published 2001 John Wiley & Sons, Inc. Int J Imaging Syst Technol, 11, 231β242, 2000