Online crop height and density estimation in grain fields using LiDAR

Abstract

Crop height and density estimation ahead of the combine harvester have been investigated over the last 30 years, but they remain a challenge. LiDAR technology is increasingly being investigated for phenotyping and monitoring of cereals. However, so far, little has been published about the influence of laser mounting position and robust online measurement of height and density from a single LiDAR scan. Therefore, the influence of the angle and height of the LiDAR mounting on crop height and density estimation in wheat and barley was investigated in this study. Tests were conducted in different crop heights, densities, moisture levels and varieties. The crop height was estimated with a root mean squared error of 0.082 m at an angle of 90° between the LiDAR scanning plane and the horizontal. Two methods were compared for crop density estimation. The inter-percentile and the transversal variance method both successfully predicted the crop density, but the variance based method performed better with a coefficient of determination (R²) of 0.77 and a root mean squared error of 82 g [dry ears] m⁻². When considering only one variety, the performance improved to reach an R² of 0.8 and a root mean squared error of 44 g [dry ears] m⁻². Variation in mounting angle of the sensor had less effect on the prediction accuracy than the mounting height.

Plant phenotyping relevance

LiDARによる作物高・密度という植物形質のオンライン推定法を開発・比較し、異なる条件で精度検証しているため、フェノタイピング手法が研究の中心である。

Code and data availability

The supplied blocks describe LiDAR data acquisition and processing for crop height/density estimation, but contain no data availability statement, no public dataset deposit, and no author code/model release. All URLs in the blocks are citations to prior work (references section), not paper-specific assets.

No evidence-backed public reproduction asset is currently recorded.