Abstract
In the Part 1 of this series, it was described how the three features of peduncle length, angle between the main stem and each peduncle and internode length could be used for the evaluation of spray formation of spray-type chrysanthemum. However, these features were measured manually. In this paper, the possibility of evaluation using image processing and neural network was investigated to automatize the process of evaluation. The evaluation indexes were calculated based on the features extracted from the processed image. From the results, the following were obtained. 1. The positions of each inflorescence and the lowest node could be detected using image processing. 2. The evaluation indexes calculated based on the features extracted from the processed image had high correlation with the evaluation value given by the expert. 3. The neural network could have less study iteration than the previous one in Part 1 of this series, since adequate evaluation indexes were used. The evaluation result of the neural network was similar enough to that of the expert.
