High throughput root phenotyping for cereal plants using spatial distribution in polar coordinate system

With the growth of the global population, the demand for better living standards and the accelerating use of grain for biofuel production, the pressures on global grain supplies are becoming immensely high. To satisfy the growing, worldwide demand for grain, it is necessary to improve the productivity of existing farmland. This is a very challenging task in the face of global environmental change. Plant breeders need to focus on traits with the greatest potential to increase yield. In this paper, we focus on the phenotyping of cereal plant roots as roots are the hidden parts of plants and are the principal organ of plants for the absorption of water and the uptake of nutrients from soil. It is well known that the root system architecture is determined by environmental factors, particularly soil conditions as well as plant's genetic makeup. This is typically called the genotype by environment (GxE) interaction. Clearly, the phenotype of a genotype is environmentally dependent. Various studies have been conducted on the effects of abiotic stresses on plant growth and adaptation, including the phenotype-genotype mapping, which has been applied to the problems from finding genes association to plant breeding. Currently, it is popular to identify the genetic factors controlling root system architecture by measuring root growth angles and there are evidences that the maximum root angles of primary roots are associated with shallow-rooted and deep-rooted cereal crops. However, there are few factors affect the results of root growth angles. The growth angle of a primary root is not a constant and the maximum growth angle may not the best trait to model the root system architecture. Instead, we propose to use spatial distribution to represent the root system architecture. In order to reduce the manual involvement to save the cost and to achieve the high-throughput and accurate root phenotyping, we develop an automated image processing software solution. In this paper, we will describe the image processing solution including automatic segmentation, non-root removal, automatic detection of top root sources, the computation of the spatial distribution of roots as well as automatic counting of root tips and the measurement of total root length.