A new complexity-based sampling optimization method is set to improve the accuracy and reliability of remote sensing interpretation in complex environments, according to a study published in the Journal of Geo-information Science.
This study proposes an optimized sampling method based on surface complexity. By calculating local complexity scores, stratifying samples using quantiles, and applying weighted sampling probabilities, this approach ensures greater sample representativeness. It effectively reduces sampling bias, enhances the accuracy of remote sensing intelligent interpretation, and provides an optimized sample selection strategy for analyzing complex surface environments. Image Credit: Beijing Zhongke Journal Publishing Co. Ltd.
Developed by researchers Lianfa Li and Xiaomei Yang from the Institute of Geographic Sciences and Natural Resources Research at the Chinese Academy of Sciences, the method enhances sample selection by incorporating surface complexity metrics, spatial heterogeneity indicators, and multi-scale morphological transformations.
Accurate sampling is crucial for remote sensing applications like land-use classification, environmental monitoring, and disaster assessment. However, traditional methods often fail to capture the full range of terrain diversity and spectral variation, leading to biased results. To address these shortcomings, the research team designed a stratified sampling approach that optimizes sample representativeness and reduces classification errors.
The study provides a comprehensive review of sampling techniques for labeled data, methods for enhancing sample diversity, and evaluation strategies for sample quality. By integrating terrain complexity with weighted stratified sampling, the method minimizes bias and improves classification accuracy. Multi-scale morphological transformations further expand sample diversity, strengthening remote sensing models.
Experimental results show that this approach significantly outperforms traditional sampling methods in capturing land surface variations and improving interpretation precision.
By improving sample representativeness through complexity-aware sampling and morphological transformations, we can enhance the reliability of remote sensing applications in challenging environments.
Lianfa Li, Professor, Spatial Statistics, State Key Laboratory of Resources & Environmental Information System, Institute of Geographic Science and Natural Resources Research, Chinese Academy of Sciences
The study offers a theoretical and technical foundation for advancing intelligent remote sensing interpretation. With applications in disaster monitoring, ecological assessment, and natural resource management, this optimized sampling strategy provides a more accurate, data-driven approach to environmental decision-making.
Journal Reference:
Li, L. et. al. (2025) Sampling Method for Complex Scene Samples in the Intelligent Interpretation of Natural Resources Remote Sensing. Journal of Geo-information Science. doi.org/10.12082/dqxxkx.2024.240278