|
|
|
| Authors: | B.S. Bennedsen, H. Moth-Poulsen |
| Keywords: | Automatic sorting, image processing, pot plant, plant quality, Pot Rose, Hibiscus |
Abstract:
The objective of the research was to establish, whether it would be possible to train an automatic system to predict the final quality of a pot plant, based on features measured on half-grown plants.
Pot roses and Hibiscus plants were used.
Images were taken of plants at a half-grown stage, the plants were allowed to grow into marketable plants, after which they were sorted by experts.
Correlation between the features, measured on the half-grown plants and the final quality were tested using linear discrimination.
For pot-roses, a reasonable correlation was found, provided the plants were classified into two classes.
For Hibiscus, the material available in this experiment proved too little to establish a correlation.
|
Download Adobe Acrobat Reader (free software to read PDF files) |
|