Non-Destructive Estimation Of Rice Leaf Area By Leaf Length And Width Measurements

The predictive regression models were used to estimate leaf area of rice with leaf length (L) and leaf width (W) measurements. Rice leaves under P (phosphorus)-deficiency and normal nutrition treatment were selected as the test samples to ensure the adaptability of the equation. Rice leaves were collected regularly in the greenhouse of Zhejiang University, Hangzhou, China. Leaf images were scanned with EPSON GT20000. The regionprops function of MATLAB was used to calculate leaf dimensions. A model (F=6.8514+0.1693L*W+0.01233(L*W)(2)-8.1261e(-5)(L*W)(3)) with L*W as the independent variable provided the most accurate estimate of leaf area (the highest R-2 (0.9493), the smallest VRMSE (3.0786) and VPRESS (2729.39) of rice leaf). The validation of the model showed that the method reproduced the measured rice leaf area well (R-2=0.9443). Measured rice leaf area was very high. Therefore, this model could be used to accurately and quickly estimate the leaf area of rice in large quantities.