Guest Editorial: Special Section on Process Imaging for Automatic Control

Industry has traditionally relied upon point sensors such as thermocouples and pressure gauges to control its operations, but as manufacturing processes become increasingly complex, additional types of information are required. For example, typical process measurement needs now include contamination detection, particulate size and shape, concentration/density profile (in pipes and tanks), and thermal profile. In many cases, imaging systems are the only sensors that can provide the required information. Process imaging is the art of visualizing events inside industrial processes. These events could be the mixing between two component materials, for example, or the completion of a chemical reaction. The image formation can be direct (e.g., acquired with a charge coupled device camera), reconstructed (e.g., tomographic imaging), or abstract (sensor data represented as an image). In advanced applications, data extracted from such images are fed back to the process control system to maintain optimum production. Process imaging encompasses well-established fields such as flow visualization and machine vision, so it can hardly be considered to be a new area. However, ongoing advances in camera and computer technology have made it feasible to apply imaging and image processing techniques to a wider range of process measurements. Of particular note is the versatility of the tomographic technology developed over the past decade. The papers in this special section highlight the current work in this area and span a wide range of technologies and applications. They were selected from the 32 papers presented at the November 2000 SPIE conference on Process Imaging for Automatic Control (Proceedings of SPIE Vol. 4188), and then underwent peer review and revision before publication in JEI. The first two papers address applications in the development of internal combustion engines. Winklhofer describes several types of applications, ranging from mapping of the flame front to monitoring air–fuel mixing behavior. Hindle et al. introduce near-infrared absorption tomography as a means of mapping the hydrocarbon distribution in the engine cylinder. In the third paper, Hall et al. demonstrate a variety of industrial uses for nuclear magnetic resonance imaging, which generates three-dimensional images. The next five papers are on the subject of electrical tomography, i.e., tomographic systems based on electrical sensing. One of the first industrial applications of such a system was in a fluidized bed at the U.S. DoE Morgantown Energy Technology Center in the mid-1980s, but during the 1990s most of the subsequent development of this technology occurred in the U.K. York presents an overview of recent industrial applications of this technology, and Mann et al. describe its application to stirred tanks. Seppanen et al. assess the ability to make more accurate estimates of multiphase flow by incorporating the prior knowledge available from fluid dynamical models; Figueroa and Seleghim evaluate two alternative sensing protocols for measuring twophase flows with electrical tomography. Mohamad-Saleh et al. show that artificial neural networks can estimate the process state directly (without first reconstructing the tomogram). The last four papers address the problem of control using multidimensional data. Deloughry et al. discuss the use of electrical capacitance tomography to control a pneumatic conveyor, and Duncan addresses the control of a general two-dimensional manufacturing operation (e.g., paper making). Sharif et al. then describe an imaging technique in which they use an array of magnetic sensors to map the transformation of the austenite phase of hot steel. Finally, Hjertaker et al. describe the control of an underwater oil/ gas separator with a linear sensor that provides a one-dimensional ‘‘image’’ of the interfaces. In making the above selections, many interesting papers from the conference have necessarily been excluded (e.g., radioactive particle tracking, microwave tomography, and principal component analysis), but they may be found in Proceedings of SPIE Volume 4188. We hope the reader will agree that process imaging does indeed provide a wide variety of tools for analyzing and controlling industrial processes. The next step is up to industry....