Maize Embryo Image Acquisition and Variety Identification Based on OTSU and K-Means Clustering Algorithm

In order to evaluate the feasibility of maize variety identification with the embryo characteristics, the paper selected four maize varieties, and scanned 70 images of each variety. It first used OSTU algorithm to segment the embryo images from the whole maize grain image. Then, it extracted six characteristic parameters of embryo from the embryo image, with connected component labeling and multi-object contour extraction algorithm. Finally, it identified the maize varieties with the six kinds of embryo´s characteristic parameters by using the k-means clustering algorithm. With these methods, the variety identification rates of the four maize varieties, including 280 test samples, are all larger than 94.12%. The experimental results demonstrate the effectiveness of maize variety identification based on embryo morphology characteristics.