Scholarlink

<p>Acquiring accurate and rapid spatial distribution of sugarcane is of great importance for agricultural production planning. Due to the cloudiness of remote sensing images, existing studies usually use 2–3&#xa0;years of multi-source data for sugarcane extraction, during which the sugarcane planting regions may change, affecting the sugarcane extraction results. In addition, existing studies have not comparatively evaluated sugarcane extraction results from commonly used crop extraction methods and do not adequately take into account sugarcane phenology information. Therefore, this study proposes the dynamic time warping method integrating sugarcane phenology characteristics and sugarcane class probabilities (SPC-DTW). Firstly, monthly maximum normalized difference vegetation index time-series data were calculated using Sentinel-2 data for 2022. Secondly, based on the training sample points, the class probabilities of sugarcane were obtained using the random forest method. Finally, the dynamic time warping method was weighted with constraints by combining the sugarcane phenology characteristics and the sugarcane class probabilities, and the 10-m sugarcane mapping result for 2022 was obtained after selecting a threshold value. The results show that the overall accuracy of the SPC-DTW method is 85.70%, and the SPC-DTW method is effective in avoiding the influence of clouds and obtaining high accuracy sugarcane mapping results.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

A Dynamic Time Warping Method Integrating Sugarcane Phenology Characteristics and Class Probabilities using Sentinel-2 NDVI Time Series

  • 期刊论文

  • 发表日期：

    2025年9月8日

查看全文
  • Linye Zhu,
  • Wenbin Sun,
  • Yiran Zhang,
  • Qi Zhang,
  • Li Zhou

摘要

Acquiring accurate and rapid spatial distribution of sugarcane is of great importance for agricultural production planning. Due to the cloudiness of remote sensing images, existing studies usually use 2–3 years of multi-source data for sugarcane extraction, during which the sugarcane planting regions may change, affecting the sugarcane extraction results. In addition, existing studies have not comparatively evaluated sugarcane extraction results from commonly used crop extraction methods and do not adequately take into account sugarcane phenology information. Therefore, this study proposes the dynamic time warping method integrating sugarcane phenology characteristics and sugarcane class probabilities (SPC-DTW). Firstly, monthly maximum normalized difference vegetation index time-series data were calculated using Sentinel-2 data for 2022. Secondly, based on the training sample points, the class probabilities of sugarcane were obtained using the random forest method. Finally, the dynamic time warping method was weighted with constraints by combining the sugarcane phenology characteristics and the sugarcane class probabilities, and the 10-m sugarcane mapping result for 2022 was obtained after selecting a threshold value. The results show that the overall accuracy of the SPC-DTW method is 85.70%, and the SPC-DTW method is effective in avoiding the influence of clouds and obtaining high accuracy sugarcane mapping results.