In plant research, there is a demand for non-destructive and non-invasive trait measurement methods for phenotyping that can be used to accurately analyze various aspects of plants, such as stem length, leaf area, and leaf inclination. In this study, a method for measuring the leaf geometric characteristics of poplar seedlings based on 3D visualization via the use of time-of-flight (ToF) and digital cameras was proposed. Firstly, the average distance density function method was applied to process outliers of leaves. Secondly, to improve the accuracy of data fitting, a specific method using the angle of adjacent two-point normal vectors was introduced to filter redundant data, kept essential sample values as the control points, and then used the control points to fit the leaf surface based on non-uniform rational B-splines (NURBS). At the same time, NURBS was used to fit the trunk according to its control points and an iterative method to fit the other branches. Finally, 3D visualization of poplar seedlings was achieved, and leaf traits, including leaf width, leaf length, leaf area, and leaf inclination angle, were calculated. To obtain accurate results, multiple experiments were conducted including assessments of poplar seedlings exhibiting normal growth and those grown under water shortage. The results of the proposed method were compared with the real values of the leaves. The RMSE for leaf width, leaf length, leaf area, and leaf angle were 0.18 cm, 0.21 cm, 1.14 cm2, and 1.97°, respectively. The results proved that this approach could be used to accurately measure the leaf characteristics of poplar seedlings via visualization. Keywords: NURBS, poplar seedling, 3D visualization, leaf geometrical characteristics DOI: 10.25165/j.ijabe.20181106.4110 Citation: Hu C H, Li P P, Pan Z. Phenotyping of poplar seedling leaves based on a 3D visualization method. Int J Agric & Biol Eng, 2018; 11(6): 145–151.
Read full abstract