一种基于多源点云数据的高精度单木分割方法
By using multi-source point cloud data fusion and physical constraint neural network inversion technology, the accuracy and efficiency problems of single tree segmentation and parameter inversion in complex forest environments have been solved, achieving high-precision extraction and identification of single tree parameters, and supporting efficient management of modern forest resource monitoring.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- SHANGHAI CHENSHAN BOTANICAL GARDEN
- Filing Date
- 2025-12-15
- Publication Date
- 2026-07-17
AI Technical Summary
Existing forest tree species identification technologies are insufficient in terms of environmental adaptability, feature fusion ability, anti-interference performance, and biological rationality. They are difficult to achieve high-precision single tree segmentation and parameter inversion, especially in complex forest environments where the identification accuracy and efficiency cannot meet the needs of modern forest monitoring.
By employing multi-source point cloud data fusion technology, combining data collected by UAV-borne LiDAR and backpack LiDAR, semantic segmentation is performed through the Enhanced Tree Instance Net model, and physical constraint neural network is introduced for parameter inversion, thereby realizing the extraction of individual tree parameters from raw point clouds in accordance with biological laws.
It improves point cloud density and data integrity, enhances the accuracy of individual tree segmentation and the biological rationality of parameter inversion, significantly improves processing efficiency and identification accuracy, and meets the needs of modern forest resource monitoring.
Smart Images
Figure CN121811031B_ABST