Estimation of the vertical leaf area profile of corn (Zea mays) plants using terrestrial laser scanning (TLS)

It remains challenging to estimate a crop's leaf area profile accurately and automatically. Terrestrial laser scanning (TLS) offers a potential solution, since it has a small footprint and can be used to rapidly and accurately resolve the canopy's 3D structure. To improve the accuracy of vertical leaf area profile estimation, we selected point clouds from three corn plants scanned in two TLS experiments and used an improved voxel-based leaf area profile estimation method based on Wilson's contact-frequency theory. The improvement has three aspects: First, it separates corn leaf points from stalk points using the difference-of-normals (DoN) method, which is commonly used to separate different surfaces of manmade objects such as tables or walls. The DoN method captures the geometrical difference between corn leaves and stalks. Second, it determines the optimal voxel size based on the calculated contact frequency. Third, it applies Wilson's voxel-based contact-frequency method to the LiDAR point clouds. Comparing the estimated and measured leaf area profiles showed that separating leaf and stalk points improved the leaf area accuracy, which depended on the voxel size. The optimal voxel sizes were 10 cm × 10 cm × 10 cm and 5 cm × 5 cm × 5 cm, respectively, in the two study years based on irradiation of the corn plant with a single laser beam. The estimated vertical and cumulative leaf area profiles agreed well with the measured profiles; mean absolute errors for the samples ranged from 15.7 to 18.9%.