Counting Maize Kernels through Digital Image Analysis

Because kernel number is the main yield component in grain crops, kernel number per plant or per unit land area is a commonly measured trait in field experiments. This is usually done by means of manual or mechanical counters, which is time consuming. We examined the possibility of counting kernels by means of image analysis. Our method involved taking photographs of dispersed kernel samples with a domestic digital camera and using the open source software ImageJ for automatic particle counting. We show that this is a quick (similar to 2 min per sample) and reliable (RMSE = 16 kernels per sample) method that could be widely adopted.