A three-stage ginseng root drying system was developed based on online image analysis. A Pilot scale batch dryer was constructed for this project. Shrinkage was monitored by computer vision system with embedded virtual controller (LabView7.0). Our experiments showed that the drying constant (kAR) calculated from the first stage, is almost equal to the drying constant (k 3) calculated from the third stage based on moisture content. This knowledge was used as feedback in order to estimate the time required to dry from 0.30 g/g (db) to 0.10 g/g (db). Temperature changes were made at critical points determined in terms of area ratio. Using an area ratio of 0.70 as the first critical point for changing air temperature resulted in changes at an average moisture content of 1.06 g/g (db) with a root mean squared error of 0.15 g/g. For the second critical point of 0.50, where the air temperature was changed back to 38°C, the end result was an average moisture content of 0.36 g/g (db), with a root mean square error of 0.067 g/g. The quality of dried ginseng roots was determined by analyzing the internal colour of roots, measured by a spectrometer. It was determined there was significant differences between big and small roots.