Journal of Taiwan Agricultural Research

Author:Hsiu-Ying Lu*, Meng-Li Wei, Chun-Tang Lu and Lit-Fu Chan

Abstract:

    As the rice trade increases, quickly and accurately distinguishing the shape differences among samples of rice grains is required to automatically separate grains for quality control. Thus, Image analysis systems become more important. The development of the software of these systems is based on the effective indices for quantifying the shape of grains. Rice grains had been traditionally classified into three categories (e.g. long, medium and short) by using the ratio of length to width, but it is too simple to adequately identify the shape variation of rice grains in image measuring system. There are few researches on the quantitative analysis of grain shape in rice. Therefore, in this study, 40 japonica rice varieties were used to determine the optimum indices for quantifying the shape in rice grain and brown rice. Geometric features such as length (L), width (W), depth (D), area (A) and perimeter (P) were measured. Seven shape parameters [two-dimensional (2D) or three-dimensional (3D)] (S1~S7) were defined and derived from these basic geometric features. Relationships among the geometric features and shape parameters were further examined. Rice grain and brown rice showed consistent results. If the 2D measurement is selected, the shape parameters S2(=L/A), S5 (=4A/πLW) and S6 (=W/L) derived from the length, width and area of grains could serve as useful indices to quantify the shape of rice grain and brown rice. If the 3D measurement is considered, the shape parameter S7(=(L+W)/2D) could be used as another shape index.

Key words:Japonica rice, Grain shape, Geometric features, Shape parameters, Image analysis system

Download: PDF Links