Rice and wheat grain counting method and software development based on Android system

Abstract

Thousand grain weight is the main component of rice and wheat yields, and an important indicator for variety breeding and cultivation management. Grain counting is an essential step for thousand grain weight measurement. Among several current counting methods, manual counting is laborious and time-consuming; electronic counting devices are expensive; the counting accuracy based on image segmentation processing is not high; and their uses are inconvenient. This study attempts to develop an application program (APP) for fast rice and wheat grain counting based on Android mobile phones for convenient use. The study identifies the relationship between image feature points and the number of grains, explores the measurement method of image feature points, and compares it with existing counting methods in terms of similarities and differences. The study also formulates grain counting calculations and develops an application program that is easy to operate. The high accuracy of this counting method has been demonstrated by tests of different varieties. The error ratio are below 2%. The program has a short running time. The counting time is generally less than one second (1s) for no more than 400 seeds. The program is convenient and easy to operate. The counting and batch-processing operations are simple. In summary, the grain counting method built in this study can be used as an effective rice and wheat grain counting tool. This study also provides a reference for the development of application programs for grain counting of other kinds of crops.