Computer-vision control system for ginseng drying

A control system for multi-stage ginseng batch drying, based on online image analysis and identification of critical control points with respect to quality, was developed. Observability was provided with a computer-vision system, which delivered relevant information about shrinkage and quality of ginseng in the batch. Shrinkage was continuously monitored through a computer-vision system by extracting of the green plane from the RGB color space followed by thresholding and pixel counting. Quality of ginseng drying was continuously monitored through computer-vision system as color intensity in the Hue-Saturation-Intensity (HSI) color space. Acceptable threshold for quality degradation was established as on the basis of 38°C drying. Controllability was provided with a predictive model of ginseng drying and a computer-vision feedback control loop. The linear relationship drying conditions and drying rate was used for online prediction of moisture content. The linear relationship between area shrinkage and moisture content was used for online estimation of actual moisture. Knowledge about significant and limiting factors of ginseng drying, obtained from the experimental study, was used for development of a pilot batch dryer. Testing of a pilot batch dryer with an embedded virtual controller (LabView7.0) proved robustness and reliability of control strategy, which provided drying to target moisture content 0.1 g/g (db) with root mean square error (RSME) of 0.0176 g/g.