A study of image measurement method for judging the growth stage of Gentiana sp. for flower harvest
2006
Shono, H.(Iwate Univ., Morioka (Japan). Faculty of Agriculture) | Touge, M.
In this study, we developed a new image-based method for judging the growth stage of Gentiana sp. for flower harvest. We automatically detected immature buds based on both outline and color information, and we then examined the utility of the method. The method is composed of two parts. The first detects buds using the generalized Hough transform based on the buds' outlines. To evaluate this technique, we designed and built experimental software. The experimental software detected nearly all of the buds, but also falsely detected a number of non-existent buds mostly on leaves. An additional processing step for discriminating between immature and non-existent buds seemed to be required for practical use of the software. Then, in the second part of the method, we select immature buds based on their color from all buds which the experimental software detected previously. The results of simulated examinations suggest that this technique could select immature buds at a very high precision. Overall, the total correct answer ratio of the overall method amounted to about 80%. It seems feasible to automatically detect immature buds. An operational realization of the method would require both additional examinations of the usefulness of the method in a wider range of situations and improved processing speed.
Показать больше [+] Меньше [-]