.png){kind=icon}
.png){kind=icon}
Return
Adaptive Neighborhood Aggregation Algorithm for PolInSAR Forest Height Estimation
.png){kind=icon}
OA
AI
DOI：10.1016/j.srs.2026.100385.png)
.png){kind=icon}
.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
Forest height is a core parameter for characterizing vertical forest structure, estimating biomass, and studying global carbon cycles. Polarized interferometric synthetic aperture radar (PolInSAR) technology possesses unique detection capabilities for vegetation vertical structure and has become an important technical means for large-scale forest height estimation. However, under single-baseline PolInSAR configurations, forest height estimation based on the random volume over ground (RVoG) model suffers from parameter solution rank deficiency. Therefore, in this paper, we propose a novel forest height estimation method based on an adaptive neighborhood aggregation algorithm (ANAA). This method overcomes the limitations of traditional fixed windows by using scattering mechanism similarity as the core metric. It quantifies differences in polarization coherence matrices using Wishart distance to dynamically construct adaptive windows, ensuring that pixels within each window satisfy the RVoG model assumptions at the physical level. Furthermore, it integrates multi-pixel observations within each window into a unified estimation framework, simultaneously solving for shared forest heights. This approach fully exploits the potential of joint observation information, fundamentally improving the rank deficiency issue. The scattering-characteristic-driven window size adaptation strategy in this article eliminates reliance on empirical window size settings. Experimental conducted in the Mabounie tropical rainforest region demonstrate that the ANAA significantly outperforms both fixed-jumping window and sliding window approaches in the key metrics of root mean square error (RMSE) and coefficient of determination (R2). RMSE is reduced by 28.8% and 20.6%, respectively, effectively enhancing estimation accuracy and robustness in complex heterogeneous forest areas. This provides an efficient and feasible solution for precise tropical rainforest forest height estimation.
Keywords:
forest height estimation
polarized interferometric SAR
RVoG model
adaptive window
joint observation of neighboring pixels
rank deficiency problem
scattering similarity
AI Summary
Key information extracted from the uploaded paper, including a brief overview, abstract, background, key highlights, visual analysis, and future outlook.
Journal
IF:
5.2
Papers:
471
Citations:
980