The Remote Sensing Ecological Index (RSEI) model is widely used for large-scale, rapid Ecological Environment Quality (EEQ) assessment. However, both the RSEI and its improved models have limitations in explaining the EEQ with only two-dimensional (2D) factors, resulting in inaccurate evaluation results. Incorporating more comprehensive, three-dimensional (3D) ecological information poses challenges for maintaining stability in large-scale monitoring, using traditional weighting methods like the Principal Component Analysis (PCA). This study introduces an Improved Remote Sensing Ecological Index (IRSEI) model that integrates 2D (normalized difference vegetation factor, normalized difference built-up and soil factor, heat factor, wetness, difference factor for air quality) and 3D (comprehensive vegetation factor) ecological factors for enhanced EEQ monitoring. The model employs a combined subjective–objective weighting approach, utilizing principal components and hierarchical analysis under minimum entropy theory. A comparative analysis of IRSEI and RSEI in Miyun, a representative study area, reveals a strong correlation and consistent monitoring trends. By incorporating air quality and 3D ecological factors, IRSEI provides a more accurate and detailed EEQ assessment, better aligning with ground truth observations from Google Earth satellite imagery.

Individual tree segmentation is crucial to extract forest structural parameters, which is vital for forest resource management and ecological monitoring. Airborne LiDAR (ALS), with its ability to rapidly and accurately acquire three-dimensional forest structural information, has become an essential tool for large-scale forest monitoring. However, accurately locating individual trees and mapping canopy boundaries continues to be hindered by the overlapping nature of the tree canopies, especially in dense forests. To address these issues, this study introduces CCD-YOLO, a novel deep learning-based network for individual tree segmentation from the ALS point cloud. The proposed approach introduces key architectural enhancements to the YOLO framework, including (1) the integration of cross residual transformer network extended (CReToNeXt) backbone for feature extraction and multi-scale feature fusion, (2) the application of the convolutional block attention module (CBAM) to emphasize tree crown features while suppressing noise, and (3) a dynamic head for adaptive multi-layer feature fusion, enhancing boundary delineation accuracy. The proposed network was trained using a newly generated individual tree segmentation (ITS) dataset collected from a dense forest. A comprehensive evaluation of the experimental results was conducted across varying forest densities, encompassing a variety of both internal and external consistency assessments. The model outperforms the commonly used watershed algorithm and commercial LiDAR 360 software, achieving the highest indices (precision, F1, and recall) in both tree crown detection and boundary segmentation stages. This study highlights the potential of CCD-YOLO as an efficient and scalable solution for addressing the critical challenges of accuracy segmentation in complex forests. In the future, we will focus on enhancing the model’s performance and application.

Estimating tree-level structural parameters from airborne laser scanning (ALS) point cloud is essential for sustainable and efficient forest management. This task is currently carried out through labour-intensive and time-consuming manual efforts, particularly in complex forest with overlapping canopies. This paper presents a novel approach for individual tree extraction utilizing an improved YOLO-based model, along with advanced algorithms for structural parameter calculation. The enhanced model, named CCD-YOLO, introduces several key improvements for individual tree segmentation (ITS) by leveraging a newly created ITS dataset. It replaces the C2F module with a CReToNeXt module to enhance feature extraction. A convolutional block attention module (CBAM) is added to highlight crown features and reduce background noise, while a Dynamic Head enables adaptive multi-layer fusion, boosting segmentation accuracy. Additionally, a denoising process is applied, ensuring more accurate and reliable measurements for vertical parameters. Lastly, an improved convex hull algorithm is employed to better accommodate the irregular shapes of tree crowns. The experimental results were evaluated in a dense forest through both internal and external consistency assessments, demonstrating significant performance enhancements. For the individual tree segmentation, the proposed approach achieved a precision of 82.4%, a recall of 72.8%, and an F1 score of 77.9%. In terms of parameter estimation, the accuracy for tree height, crown width, and crown area was 0.82, 0.60, and 0.99, respectively, with corresponding RMSE values of 1.25, 1.01, and 1.16. These results highlight the effectiveness of the proposed method in improving both segmentation and parameter estimation accuracy.