Development of Precision Production Robot for Flower and Vegetable Seedling (Part 1)

Discrimination of Stock Seedling by Image Processing

Abstract

In the process of producing flower and vegetable seedlings, defective ones are produced by being germinally inhibited, or being effected by inappropriate environment, moreover, some sort of diseases and insect damages. In order to improve the seedling production, a precision production robot, which removes defective seedlings and transplants good ones into cell pots, is necessary. For this robot, we accumulated picture property of seedlings by image processing. In this paper, we focused on the discrimination method of stock seedlings, which is thought to be most difficult so that skilled workers are necessary. The robot could be applied to most flower and vegetable seedlings if this discrimination of stock seedlings is possible. Then, we firstly made the algorithm of discrimination between double and single flowering forms in the seedling stage, which is based on picture property by image processing. In the result, the accuracy of eight folds seedling discrimination using properties (area, complexity degree and color of cotyledon etc.) was about 89%.