An automated method for the assessment of the rice grain germination rate

Abstract

The germination rate of rice grain is recognized as one of the most significant indicators of seed quality assessment. Currently, grain germination rate is generally determined manually by experienced researchers, which is time-consuming and labor-intensive. In this paper, a new method is proposed for counting the number of grains and germinated grains. In the coarse segmentation process, the k-means clustering algorithm is applied to obtain rough grain-connected regions. We further refine the segmentation results obtained by the k-means algorithm using a one-dimensional Gaussian filter and a fifth-degree polynomial. Next, the optimal single grain area is determined based on the area distribution curve. Accordingly, the number of grains contained in the connected region is equal to the area of the connected region divided by the optimal single grain area. Finally, a novel algorithm is proposed for counting germinated grains. This algorithm is based on the idea that the length of the intersection between the germ and the grain is less than the circumference of the germ. The experimental results show that the mean absolute error of the proposed method for germination rate is 2.7%. And the performance of the proposed method is robust to changes in grain number, grain varieties, scale, illumination, and rotation.