Separating Tree Photosynthetic and Non-Photosynthetic Components from Point Cloud Data Using Dynamic Segment Merging

Abstract

Many biophysical forest properties such as wood volume and leaf area index (LAI) require prior knowledge on either photosynthetic or non-photosynthetic components. Laser scanning appears to be a helpful technique in nondestructively quantifying forest structures, as it can acquire an accurate three-dimensional point cloud of objects. In this study, we propose an unsupervised geometry-based method named Dynamic Segment Merging (DSM) to identify non-photosynthetic components of trees by semantically segmenting tree point clouds, and examining the linear shape prior of each resulting segment. We tested our method using one single tree dataset and four plot-level datasets, and compared our results to a supervised machine learning method. We further demonstrated that by using an optimal neighborhood selection method that involves multi-scale analysis, the results were improved. Our results showed that the overall accuracy ranged from 81.8% to 92.0% with an average value of 87.7%. The supervised machine learning method had an average overall accuracy of 86.4% for all datasets, on account of a collection of manually delineated representative training data. Our study indicates that separating tree photosynthetic and non-photosynthetic components from laser scanning data can be achieved in a fully unsupervised manner without the need of training data and user intervention.

LeWoS: A universal leaf‐wood classification method to facilitate the 3D modelling of large tropical trees using terrestrial LiDAR

“…Strom, Richardson, & Olson, ), and exercise it recursively in order to achieve a robust segmentation. Moreover, we alter direct wood extraction, which requires manual fine‐tuning of some thresholds (Wang et al, ), to a class probability estimation. This transformation not only greatly increases the universality of the proposed LeWoS method, but also provides classification confidence information.…”

“…In the previous decade, some efforts have been made to separate leaf and wood components in TLS point clouds (Béland, Baldocchi, Widlowski, Fournier, & Verstraete, ; Hackenberg, Spiecker, Calders, Disney, & Raumonen, ; Hétroy‐Wheeler, Casella, & Boltcheva, ; Ma et al, ; Vicari, Disney, et al, ; Wang et al, ). However, separating these components in TLS data still remains a main challenge, especially in evergreen tropical forests.…”

“…Overall, existing methods were either based on geometric (Wang et al, ) or radiometric intensity features (Béland et al, ) or a combination of both (Zhu et al, ). Methods based on intensity usually have limited application scenarios.…”

“…Nevertheless, such properties are generally influenced by the distance, a partial laser hit and its incidence angle (Kaasalainen, Krooks, Kukko, & Kaartinen, ). Intensity values have to be calibrated for each specific instrument (Calders et al, ), which limits the applicability of using intensity information for subsequent processing (Wang et al, ). Some research‐led developments on multi‐wavelength systems may help to differentiate leaf and wood in the future (Disney, ).…”

“…These unsupervised methods often look at specific geometric properties of components. Examples include presuming leaves being flat (Li et al, ), trunk and branch boundaries appear as circles or circle‐like shapes (Tao et al, ), with linear structures across scales (Wang et al, ).…”