Fundamental study for estimating rice-plant stem number using laser scanner measurements

Abstract

The number of rice-plant stems (S), directly affecting the competition among rice plants and contributing to rice yield, is estimated from laser data. The laser data were normalized to eliminate the increasing plant height effect. Relative spatial volume (rV laser s) was derived and scaled as an exponential function of S. A relationship between rV laser s and S is confirmed in two growing seasons (2014 and 2016, separately obtained by two different laser scanners). The scaling and exponent factors (β and α) depended on the planting geometry, planting density, and bottom position of the rice plants (D bottom) but were almost independent of the number of divided layers in the rV laser s computation. From the estimated stem number using D bottom at the 80th percentile (D 80), the maximum S was obtained at ∼50 days after transplanting. In both years, the relative error in the estimate was below 0.10, and the bias was small. In the models with D 80 in 2014 (MD 80 2014) and D 95 in 2016 (MD 95 2016), the β and α values were very similar. Using the rV laser s measure, we can disregard the footprint characteristics and voxel size. The presented results support the proposed approach as a useful future method for estimating rice-plant stems. © The Authors. Published by SPIE under a Creative Commons Attribution 3.0 Unported License. Distribution or reproduction of this work in whole or in part requires full attribution of the original publication, including its DOI.