In the present circumstances, growers of
Gentiana sp., which is the representative flowering plant cultivated at Iwate prefecture in Japan, usually judge harvest stage with their own eyes subjectively. This way of judging often results in misjudgments and destabilizes the quality of shipped flowers. So, in this study, we examined the availability of image-based information to provide objective indices for proper judgement of the growth stage of
Gentiana sp. for flower harvest.
First of all, we defined 2 groups of image-based information. The first group contains 4 kinds of information based on the area of petals and buds in an image, and the second group contains 4 kinds of information based on number of buds in an image.
Next, we took a total of 58 photos of
Gentiana in the field and calculated the proposed information from these photos. Then, we checked relationship between the proposed information and growth stages as evaluated by an expert in growth stage judgment.
The 6 kinds of information showed considerable correlation with the growth stages as evaluated by the expert.
To more precisely estimate the growth stage, we tried to combine information from the proposed information using multiple linear regression analysis. The correlation coefficient between the combined information and the growth stages evaluated by the expert was beyond 0.9 in 2 verifications.
Thus, it seems that the proposed information has possibility to stabilize the quality of shipped flowers by providing objective indices for the proper judgement of growth stage of
Gentiana sp., and therefore has capability to contribute to automation of
Gentiana sp. flower harvest in the future.
View full abstract