A LiDAR Point Cloud Column Projection Method for Estimating Zone-Specific View-Dependent Canopy Gap Fraction of Fruit Tree Canopies

Abstract

View-dependent canopy gap fraction is a directional descriptor of canopy openness that provides structural information for characterizing fruit-tree canopy heterogeneity. However, existing LiDAR-based canopy characterization methods are limited in their stability for characterizing zone-specific variations in canopy occlusion. This study proposes a LiDAR point-cloud column projection method for estimating zone-specific view-dependent canopy gap fraction of fruit tree canopies. A flash LiDAR sensor and an RGB camera were used to acquire paired point-cloud and image data from a simulated fruit tree and real peach trees under matched viewing directions and consistent region-of-interest constraints. Image-derived canopy gap fraction, calculated from the blue-cloth background area in each 3 × 3 subregion, was used as the reference value. The proposed column projection method was compared with a regional volumetric occupancy method and a regional two-dimensional projection coverage method. Under the indoor method-development conditions, using 88 paired frames, the proposed method achieved a mean absolute error of 8.59%, a Pearson correlation coefficient of 0.925, and a Spearman rank correlation coefficient of 0.914 relative to the image-derived reference values. In the preliminary semi-controlled outdoor evaluation using eight paired real-tree samples, the method achieved a mean absolute error of 10.48% and showed lower numerical errors and systematic bias than the two baseline methods. These results indicate that the proposed method can provide zone-specific view-dependent canopy gap fraction estimates that are more consistent with image-derived reference values under the tested conditions.