Quantifying Pesticide Deposition on Apple Surfaces Based on Multi-View Tracer Point Cloud Reconstruction: Laboratory Characterization and Orchard Assessment

To address the limitations of conventional point-based sampling methods (e.g., water-sensitive paper and filter paper) in characterizing the three-dimensional deposition distribution on curved fruit surfaces, this study proposed a fruit-surface deposition quantification approach validated through a coupled workflow comprising multi-view 3D reconstruction, target point cloud segmentation, and chemical wash-off quantification. A carmine solution was used as a tracer to simulate pesticide deposition. High-accuracy colored point clouds of apples were generated via multi-view 3D reconstruction, and the target deposition point cloud was automatically extracted using a combined strategy integrating K-means-based coarse color clustering with HSV threshold refinement. The target point cloud count was regressed against the measured single-fruit deposition volume. The results showed a significant linear relationship between the target point cloud count and deposition under both laboratory and real orchard air-assisted spraying conditions (laboratory: R2 = 0.827, r = 0.909, n = 13, p < 0.01; orchard: R2 = 0.885, r = 0.941, n = 17, p < 0.01). This method can characterize the differences in pesticide deposition among fruits while preserving the three-dimensional location information of the tracer-related region, thus providing a methodological basis for characterizing pesticide deposition on the fruit surface.