Maize Variety Identification Based on Support Vector Machines

Abstract

Seeds are the most basic and critical means of production in agricultural production. Declining purity of seed varieties will significantly reduce crop yield and product quality. The quality of crop seed varieties, especially purity testing, can protect the interests of farmers and promote the development of rural economies. The aim of this paper is to study maize variety identification based on support vector machines. Based on the morphological structure characteristics of maize varieties, a set of characteristic parameters that can accurately reflect the morphological structure of different maize varieties is proposed for variety identification. The parameter set consists of four components: shape, colour, size and sharpness parameters. The experimental results show that the support vector machine algorithm outperforms the BP neural network.