Abstract
In the previous papers of this series, the spray formation of cut chrysanthemum by the use of image processing and neural network was evaluated and the usefulness of the method was reported. However, cut chrysanthemums used were without leaves. In this paper, the automatic evaluation system for spray type chrysanthemum with leaves was examined. First, the spray formation of the chrysanthemum was expressed by polygonal approximation in the procedure of image processing. Secondly, the evaluation indexes were calculated based on the polygon. Thirdly, theevaluation indexes were inputted to neural networks and some cut chrysanthemums were evaluated. The following results were obtained : 1. Binary images of whole chrysanthemum and inflorescence were appropriately extracted from the original image through two threshold levels calculated by the gray level histogram. From the binary image, position of the bottom node of the chrysanthemum was precisely detected. 2. The shapes of the spray formations were approximated by polygons in the binary images. 3. The three evaluation indexes, which were related to the approximate shape of the cut chrysanthemum and the degree of dispersion of inflorescence, were calculated based on the polygons. 4. The evaluations by neural networks with three indexes corresponded to those by experts.
