抄録
In the intentional production of crops grown in a plant factory, it is necessary to measure the progress of growth of an individual plant. The image processing of plant growth is an effective tool for intact and non-destructive measurement, so as to be able to digitize continuously the images of plant or leaves with the computer system. There is a significant positive correlation between several growth indices obtained by image processing and the top fresh weight of lettuce plants (Lactuca sativa L. non-heading type) grown in a plant factory. The correlation coefficient between the horizontal projected leaf area and the top fresh weight was highest among the image features of lettuce, i.e. 0.947 or higher. During the early growth period from two-to eight-leaf stage, the correlation coefficient between the horizontal projected leaf area and the top fresh weight was higher than that of the whole span of lettuce's life, i.e. 0.95 or higher. The horizontal projected leaf area for several days after sowing made it possible to estimate the top fresh weight. And so, we have developed an automatic growth measurement system by image processing. To measure individual plants continuously and automatically, we have developed the algorithms for image processing and measurement. The algorithms were composed of the labelling for recognizing individual plant, calculating of argument suited to the growth process and automatic calibration for eliminating the error caused by plant age. From these, the top fresh weight of individual plants was obtained continuously by use of a low-cost automatic plant growth measuring system with a personal computer.
