Non-destructive identification of pure breeding Rice seed using digital image analysis

Abstract

In this paper, digital image analysis is applied for non-destructive identification of pure breeding Rice seed. The appearance of rice such as its shape and color is expected to be the important features in agricultural breeding and quality testing. It is a difficult task for farmer to identify rice seeds because of the similar surface color of the seeds. This paper presents an automatic classification method based on segment images and RGB color features. Hardware of image capturing is designed using scanner. The ratio between segment images and varieties of different shades RGB histogram are then calculated. The rule of classification “Khao Dawk Mali 105” between pure breeding Rice seed and impure breeding Rice seed are created. The correct classification rates for two steps are: good rice seeds 98% and pure breeding rice seeds 82%. This information could be used as a signal to farmer decided to switch to a new generation seeds.