1. NCCU Academic Hub
  2. Publications
  3. College of Commerce
  4. Theses
  5. Item

Publications-Theses

Article View/Open

Publication Export

Google ScholarTM

NCCU Library

Citation Infomation

Related Publications in TAIR

題名 利用結構樹狀幾何進行基於衛星影像時間序列的作物分類
Classification via structural tree-geometry upon crop-classes of satellite image time series
作者 林兆鵬
Lin, Jhao-Peng
貢獻者 周珮婷<br>謝復興
林兆鵬
Lin, Jhao-Peng
關鍵詞 時間序列分類
樹狀幾何結構
階層式分群
多尺度分類
衛星影像
相互資訊
Time-series classification
Tree geometry
Hierarchical clustering
Multiscale classification
Satellite imagery
Mutual information
日期 2025
上傳時間 4-Aug-2025 15:10:57 (UTC+8)
摘要 對眾多時間序列片段進行分類是一項在計算上與概念上都極具挑戰性的任務。類別間未知的相似程度會降低分類準確率，而對時間序列本質的認知不足，則使得我們無從判斷關鍵時間區段。由福爾摩沙二號衛星影像衍生出的「作物時間序列資料集」，包含 24 類衛星影像產生的時間序列，每條序列在時間軸上共有 46 個觀測點，同時呈現了上述兩個分類上的挑戰。為了解決這些問題，我們針對這 24 個作物類別構建一種樹狀幾何結構，以利於利用不同分支（或稱「屬」）間在訊號與雜訊比上的差異性，並將同一屬內的分類簡化為較小的子任務。本研究提出並實作了一種多尺度的屬間分類方法。該方法首先根據時間軸上，一階與二階差分的連續關聯性所選出的特徵片段定義距離，並據此對時間序列進行階層式分群，以捕捉時間序列的功能性模式。分類流是依據階層式分群樹的內部節點進行分支，當通過其可靠性檢驗時便進一步細分。對於每一次屬間分類，皆建立一張熱圖以呈現分類效果，同時也會展示一組不涉入該分類的屬所對應的熱圖，以揭示潛在的離群資訊。透過此方式即構建出一個多尺度的分類流程。
It is a computational and conceptual challenge when making classifications among many classes of time-series segments. Unknown degrees of similarity among classes drive the classification accuracy low, while lacking knowledge underlying time series causes not knowing where important temporal regions are and where they are not. The Crop time series data set derived from Formosa-2 satellite images with 24 classes of satellite-image-induced time series on an axis of 46 temporal-points simultaneously presents these two challenging aspects onto the classification task. To resolve this task, a tree-geometry is computed upon the 24 crop-classes to facilitate exploitation of differential signal-to-noise (S-N) ratios across all between-branch (or genus) classifications and reduce all within-genus classifications into minor tasks. A multiscale between-genus classification methodology is proposed and implemented by first constructing a Hierarchical clustering (HC) tree on time series with a distance defined by motifs selected according to serial 1st- and 2nd-order difference-based associations along the temporal axis to capture the time series' functional patterns. The classification proceeds by splitting branches at internal nodes of HC-tree whenever passing its reliability check. Via a heatmap built for each between-genus classification, its classification efficiency is manifested and simultaneously a collection of non-involving genus-based heatmaps is shown to shed light on outlier information. As such a multiscale classification protocol is built.
參考文獻 Anderson, P. W. (1972). More is different: Broken symmetry and the nature of the hierarchical structure of science. Science, 177(4047), 393–396. Antonijević, O., Jelić, S., Bajat, B., & Kilibarda, M. (2023). Transfer learning approach based on satellite image time series for the crop classification problem. Journal of Big Data, 10(1), 54. Avcı, M. (2000). A hierarchical classification of landsat tm imagery for land cover mapping (Unpublished master’s thesis). Middle East Technical University. Belgiu, M., & Csillik, O. (2018). Sentinel-2 cropland mapping using pixel-based and object based time-weighted dynamic time warping analysis. Remote sensing of environment, 204, 509–523. Buza, K.,&Schmidt-Thieme, L. (2010). Motif-basedclassification of time series with bayesian networks and svms. In Advances in data analysis, data handling and business intelligence: Proceedings of the 32nd annual conference of the gesellschaft für klassifikation ev, joint conference with the british classification society (bcs) and the dutch/flemish classification society (voc), helmut-schmidt-university, hamburg, july 16-18, 2008 (pp. 105–114). Cover, T. M. (1999). Elements of information theory. John Wiley & Sons. Dau, H. A., Keogh, E., Kamgar, K., Yeh, C.-C. M., Zhu, Y., Gharghabi, S., … Hexagon-ML (2018, October). The ucr time series classification archive. (https://www.cs.ucr.edu/~eamonn/time_series_data_2018/) Del Moral, P., Nowaczyk, S., Sant’Anna, A., & Pashami, S. (2023). Pitfalls of assessing extracted hierarchies for multi-class classification. Pattern Recognition, 136, 109225. Fushing, H., Chou, E. P., & Chen, T.-L. (2023). Multiscale major factor selections for complex system data with structural dependency and heterogeneity. Physica A: Statistical Mechanics and its Applications, 630, 129227. Fushing, H., Kao, H.-W., & Chou, E. P. (2024). Topological risk-landscape in metric-free categorical database. IEEE Access. Gell-Mann, M. (2002). What is complexity? In A. Q. Curzio & M. Fortis (Eds.), Complexity and industrial clusters (pp. 13–24). Heidelberg: Physica-Verlag HD. Largouët, C., & Cordier, M.-O. (2001). Improving the landcover classification using domain knowledge. AI Communications, 14(1), 35–43. Nidamanuri, R. R., & Zbell, B. (2012). Existence of characteristic spectral signatures for agri cultural crops–potential for automated crop mapping by hyperspectral imaging. Geocarto International, 27(2), 103–118. Pal, M. (2005). Randomforestclassifier for remotesensing classification. International journal of remote sensing, 26(1), 217–222. Pelletier, C., Webb, G. I., & Petitjean, F. (2019). Temporal convolutional neural network for the classification of satellite image time series. Remote Sensing, 11(5), 523. Petitjean, F., Inglada, J., & Gançarski, P. (2012). Satellite image time series analysis under time warping. IEEE transactions on geoscience and remote sensing, 50(8), 3081–3095. Radoi, A. (2022). Multimodal satellite image time series analysis using gan-based domain translation and matrix profile. Remote Sensing, 14(15), 3734. Reichstein, M., Camps-Valls, G., Stevens, B., Jung, M., Denzler, J., Carvalhais, N., & Prabhat, F. (2019). Deep learning and process understanding for data-driven earth system science. Nature, 566(7743), 195–204. Rußwurm, M., & Körner, M. (2018). Multi-temporal land cover classification with sequential recurrent encoders. ISPRS International Journal of Geo-Information, 7(4), 129. Shiu, S.-Y., Chin, Y.-S., Lin, S.-H., & Chen, T.-L. (2024). Randomized self-updating process for clustering large-scale data. Statistics and Computing, 34(1), 47. Sneath, P. H. (2005). Numerical taxonomy. In Bergey’s manual® of systematic bacteriology (pp. 39–42). Springer. Tan, C. W., Webb, G. I., & Petitjean, F. (2017). Indexing and classifying gigabytes of time series under time warping. In Proceedings of the 2017 siam international conference on data mining (pp. 282–290). Tumer, K., & Wolpert, D. H. (2004). Collectives and the design of complex systems. Springer Science & Business Media. Wu, B., Zhang, M., Zeng, H., Tian, F., Potgieter, A. B., Qin, X., … others (2023). Challenges and opportunities in remote sensing-based crop monitoring: A review. National Science Review, 10(4), nwac290. Yang, J., Gong, P., Fu, R., Zhang, M., Chen, J., Liang, S., … Dickinson, R. (2013). The role of satellite remote sensing in climate change studies. Nature climate change, 3(10), 875–883. Zhang, T., Cheng, C., & Wu, X. (2023). Mapping the spatial heterogeneity of global land use and land cover from 2020 to 2100 at a 1 km resolution. Scientific Data, 10(1), 748. Zhao, Q., Yu, L., Du, Z., Peng, D., Hao, P., Zhang, Y., & Gong, P. (2022). An overview of the applications of earth observation satellite data: impacts and future trends. Remote Sensing, 14(8), 1863. Zhu, X. X., Tuia, D., Mou, L., Xia, G.-S., Zhang, L., Xu, F., & Fraundorfer, F. (2017). Deep learning in remote sensing: A comprehensive review and list of resources. IEEE geo-science and remote sensing magazine, 5(4), 8–36.
描述 碩士
國立政治大學
統計學系
112354001
資料來源 http://thesis.lib.nccu.edu.tw/record/#G0112354001
資料類型 thesis
dc.contributor.advisor 周珮婷<br>謝復興zh_TW
dc.contributor.author (Authors) 林兆鵬zh_TW
dc.contributor.author (Authors) Lin, Jhao-Pengen_US
dc.creator (作者) 林兆鵬zh_TW
dc.creator (作者) Lin, Jhao-Pengen_US
dc.date (日期) 2025en_US
dc.date.accessioned 4-Aug-2025 15:10:57 (UTC+8)-
dc.date.available 4-Aug-2025 15:10:57 (UTC+8)-
dc.date.issued (上傳時間) 4-Aug-2025 15:10:57 (UTC+8)-
dc.identifier (Other Identifiers) G0112354001en_US
dc.identifier.uri (URI) https://nccur.lib.nccu.edu.tw/handle/140.119/158711-
dc.description (描述) 碩士zh_TW
dc.description (描述) 國立政治大學zh_TW
dc.description (描述) 統計學系zh_TW
dc.description (描述) 112354001zh_TW
dc.description.abstract (摘要) 對眾多時間序列片段進行分類是一項在計算上與概念上都極具挑戰性的任務。類別間未知的相似程度會降低分類準確率，而對時間序列本質的認知不足，則使得我們無從判斷關鍵時間區段。由福爾摩沙二號衛星影像衍生出的「作物時間序列資料集」，包含 24 類衛星影像產生的時間序列，每條序列在時間軸上共有 46 個觀測點，同時呈現了上述兩個分類上的挑戰。為了解決這些問題，我們針對這 24 個作物類別構建一種樹狀幾何結構，以利於利用不同分支（或稱「屬」）間在訊號與雜訊比上的差異性，並將同一屬內的分類簡化為較小的子任務。本研究提出並實作了一種多尺度的屬間分類方法。該方法首先根據時間軸上，一階與二階差分的連續關聯性所選出的特徵片段定義距離，並據此對時間序列進行階層式分群，以捕捉時間序列的功能性模式。分類流是依據階層式分群樹的內部節點進行分支，當通過其可靠性檢驗時便進一步細分。對於每一次屬間分類，皆建立一張熱圖以呈現分類效果，同時也會展示一組不涉入該分類的屬所對應的熱圖，以揭示潛在的離群資訊。透過此方式即構建出一個多尺度的分類流程。zh_TW
dc.description.abstract (摘要) It is a computational and conceptual challenge when making classifications among many classes of time-series segments. Unknown degrees of similarity among classes drive the classification accuracy low, while lacking knowledge underlying time series causes not knowing where important temporal regions are and where they are not. The Crop time series data set derived from Formosa-2 satellite images with 24 classes of satellite-image-induced time series on an axis of 46 temporal-points simultaneously presents these two challenging aspects onto the classification task. To resolve this task, a tree-geometry is computed upon the 24 crop-classes to facilitate exploitation of differential signal-to-noise (S-N) ratios across all between-branch (or genus) classifications and reduce all within-genus classifications into minor tasks. A multiscale between-genus classification methodology is proposed and implemented by first constructing a Hierarchical clustering (HC) tree on time series with a distance defined by motifs selected according to serial 1st- and 2nd-order difference-based associations along the temporal axis to capture the time series' functional patterns. The classification proceeds by splitting branches at internal nodes of HC-tree whenever passing its reliability check. Via a heatmap built for each between-genus classification, its classification efficiency is manifested and simultaneously a collection of non-involving genus-based heatmaps is shown to shed light on outlier information. As such a multiscale classification protocol is built.en_US
dc.description.tableofcontents 第壹章 緒論 1 第一節 研究動機與目的 1 第二節 資料介紹 5 第貳章 文獻探討 7 第參章 科學資料分析觀點下的分類 9 第一節 類別資料的多樣性與曲線模式形成 9 第二節 分類體系中的階層結構 10 第三節 全資訊導向的分析 11 第肆章 研究方法 12 第一節 Randomized self-updating process 13 第二節 曲線特徵片段（motif）之建構 21 第三節 兩兩屬別間的分類與可靠性檢驗方法 24 第一項 兩兩屬別分類流程：以 7, 8, 13, 16 與 2, 17, 19 為例 24 第二項 兩兩屬別比較與進一步分裂的可靠性檢驗 28 第三項 離群點偵測：以類別為基礎的全域視角 32 第伍章 資料分析結果 35 第一節 比較 7, 8, 13, 16 與 2, 17, 19 兩個屬 35 第二節 比較 7, 8, 13, 16 與 1, 10, 15 兩個屬 36 第三節 比較 7, 8, 13, 16 與 14, 18, 20 兩個屬 37 第四節 比較 7, 8, 13, 16 與 6, 9, 12 兩個屬 39 第五節 比較 6, 9, 12 與 1, 10, 15 兩個屬 41 第六節 屬內分類：6, 9, 12 43 第七節 屬內分類：14, 18, 20 43 第八節 屬內分類：1, 10, 15 43 第九節 屬內分類：7, 8, 13, 16 43 第十節 屬內分類：2, 17, 19 48 第十一節 屬內分類：3, 4, 5, 22 48 第陸章 結論與建議 51 參考文獻 53zh_TW
dc.format.extent 31166805 bytes-
dc.format.mimetype application/pdf-
dc.source.uri (資料來源) http://thesis.lib.nccu.edu.tw/record/#G0112354001en_US
dc.subject (關鍵詞) 時間序列分類zh_TW
dc.subject (關鍵詞) 樹狀幾何結構zh_TW
dc.subject (關鍵詞) 階層式分群zh_TW
dc.subject (關鍵詞) 多尺度分類zh_TW
dc.subject (關鍵詞) 衛星影像zh_TW
dc.subject (關鍵詞) 相互資訊zh_TW
dc.subject (關鍵詞) Time-series classificationen_US
dc.subject (關鍵詞) Tree geometryen_US
dc.subject (關鍵詞) Hierarchical clusteringen_US
dc.subject (關鍵詞) Multiscale classificationen_US
dc.subject (關鍵詞) Satellite imageryen_US
dc.subject (關鍵詞) Mutual informationen_US
dc.title (題名) 利用結構樹狀幾何進行基於衛星影像時間序列的作物分類zh_TW
dc.title (題名) Classification via structural tree-geometry upon crop-classes of satellite image time seriesen_US
dc.type (資料類型) thesisen_US
dc.relation.reference (參考文獻) Anderson, P. W. (1972). More is different: Broken symmetry and the nature of the hierarchical structure of science. Science, 177(4047), 393–396. Antonijević, O., Jelić, S., Bajat, B., & Kilibarda, M. (2023). Transfer learning approach based on satellite image time series for the crop classification problem. Journal of Big Data, 10(1), 54. Avcı, M. (2000). A hierarchical classification of landsat tm imagery for land cover mapping (Unpublished master’s thesis). Middle East Technical University. Belgiu, M., & Csillik, O. (2018). Sentinel-2 cropland mapping using pixel-based and object based time-weighted dynamic time warping analysis. Remote sensing of environment, 204, 509–523. Buza, K.,&Schmidt-Thieme, L. (2010). Motif-basedclassification of time series with bayesian networks and svms. In Advances in data analysis, data handling and business intelligence: Proceedings of the 32nd annual conference of the gesellschaft für klassifikation ev, joint conference with the british classification society (bcs) and the dutch/flemish classification society (voc), helmut-schmidt-university, hamburg, july 16-18, 2008 (pp. 105–114). Cover, T. M. (1999). Elements of information theory. John Wiley & Sons. Dau, H. A., Keogh, E., Kamgar, K., Yeh, C.-C. M., Zhu, Y., Gharghabi, S., … Hexagon-ML (2018, October). The ucr time series classification archive. (https://www.cs.ucr.edu/~eamonn/time_series_data_2018/) Del Moral, P., Nowaczyk, S., Sant’Anna, A., & Pashami, S. (2023). Pitfalls of assessing extracted hierarchies for multi-class classification. Pattern Recognition, 136, 109225. Fushing, H., Chou, E. P., & Chen, T.-L. (2023). Multiscale major factor selections for complex system data with structural dependency and heterogeneity. Physica A: Statistical Mechanics and its Applications, 630, 129227. Fushing, H., Kao, H.-W., & Chou, E. P. (2024). Topological risk-landscape in metric-free categorical database. IEEE Access. Gell-Mann, M. (2002). What is complexity? In A. Q. Curzio & M. Fortis (Eds.), Complexity and industrial clusters (pp. 13–24). Heidelberg: Physica-Verlag HD. Largouët, C., & Cordier, M.-O. (2001). Improving the landcover classification using domain knowledge. AI Communications, 14(1), 35–43. Nidamanuri, R. R., & Zbell, B. (2012). Existence of characteristic spectral signatures for agri cultural crops–potential for automated crop mapping by hyperspectral imaging. Geocarto International, 27(2), 103–118. Pal, M. (2005). Randomforestclassifier for remotesensing classification. International journal of remote sensing, 26(1), 217–222. Pelletier, C., Webb, G. I., & Petitjean, F. (2019). Temporal convolutional neural network for the classification of satellite image time series. Remote Sensing, 11(5), 523. Petitjean, F., Inglada, J., & Gançarski, P. (2012). Satellite image time series analysis under time warping. IEEE transactions on geoscience and remote sensing, 50(8), 3081–3095. Radoi, A. (2022). Multimodal satellite image time series analysis using gan-based domain translation and matrix profile. Remote Sensing, 14(15), 3734. Reichstein, M., Camps-Valls, G., Stevens, B., Jung, M., Denzler, J., Carvalhais, N., & Prabhat, F. (2019). Deep learning and process understanding for data-driven earth system science. Nature, 566(7743), 195–204. Rußwurm, M., & Körner, M. (2018). Multi-temporal land cover classification with sequential recurrent encoders. ISPRS International Journal of Geo-Information, 7(4), 129. Shiu, S.-Y., Chin, Y.-S., Lin, S.-H., & Chen, T.-L. (2024). Randomized self-updating process for clustering large-scale data. Statistics and Computing, 34(1), 47. Sneath, P. H. (2005). Numerical taxonomy. In Bergey’s manual® of systematic bacteriology (pp. 39–42). Springer. Tan, C. W., Webb, G. I., & Petitjean, F. (2017). Indexing and classifying gigabytes of time series under time warping. In Proceedings of the 2017 siam international conference on data mining (pp. 282–290). Tumer, K., & Wolpert, D. H. (2004). Collectives and the design of complex systems. Springer Science & Business Media. Wu, B., Zhang, M., Zeng, H., Tian, F., Potgieter, A. B., Qin, X., … others (2023). Challenges and opportunities in remote sensing-based crop monitoring: A review. National Science Review, 10(4), nwac290. Yang, J., Gong, P., Fu, R., Zhang, M., Chen, J., Liang, S., … Dickinson, R. (2013). The role of satellite remote sensing in climate change studies. Nature climate change, 3(10), 875–883. Zhang, T., Cheng, C., & Wu, X. (2023). Mapping the spatial heterogeneity of global land use and land cover from 2020 to 2100 at a 1 km resolution. Scientific Data, 10(1), 748. Zhao, Q., Yu, L., Du, Z., Peng, D., Hao, P., Zhang, Y., & Gong, P. (2022). An overview of the applications of earth observation satellite data: impacts and future trends. Remote Sensing, 14(8), 1863. Zhu, X. X., Tuia, D., Mou, L., Xia, G.-S., Zhang, L., Xu, F., & Fraundorfer, F. (2017). Deep learning in remote sensing: A comprehensive review and list of resources. IEEE geo-science and remote sensing magazine, 5(4), 8–36.zh_TW