潮水線影像辨識法建構潮間帶及海岸濕地數值高程地圖

Enhancing waterline image interpretation to construct digital elevation maps of estuarine and coastal wetlands

徐偉銓(Wei-Chuan Hsu)

指導教授 ： 施上粟

國立臺灣大學/工學院/土木工程學研究所/碩士(2021年)

https://doi.org/10.6342/NTU202104144

摘要

濕地為全球生產力最豐富的生態系之一，其中海岸濕地是指位於河海交流處並受到潮汐影響的局部封閉區域，包含了紅樹林、海草床、珊瑚礁等珍貴的藍碳生態系統。其中地形高程資料是濕地經營管理上重要的關鍵資訊，因此快速有效地獲取海岸濕地的地形高程資料將有助於自然資源的經營管理。然而，目前的地形量測技術除了因為需投入較大量的人力及經費，而難以快速反覆地取得擷取地形資料外，也缺乏透過空照圖資追溯建立歷史地形的關鍵技術。本研究假設研究區域的水位高度為相同且平坦的前提下，提出以水線(Waterline)結合潮汐水位資料以快速地建構海岸濕地數值高程地圖的方法，因此如何正確地辨識水線資訊的時間及空間變化是本研究的主要挑戰。為了確立最佳的水線辨識流程，本研究利用相機取得前導實驗模擬現地水體與灘地交界的影像，並透過施放tracer 染劑及影像強化(Image enhancement)分別達到模擬現地水色深淺及增加整體影像對比度的效果。本研究比較傳統像元式分類法(Pixel-based Classification)，及近年常見的物件式導向分析技術(Object-Based Image Analysis ; OBIA)對影像的判釋差異，也透過雜訊濾除及數學形態學的運算處理提升辨識水線位置的精度，並利用邊緣檢測(Edge Detection)技術以截取水線位置。實驗室的前導實驗研究結果發現，透過水體染色能夠提升大多數情況的水線的辨識精度，而將原正射影像進行強化處理能夠有效提升OBIA得到的水線辨識精度。另外，本研究亦利用無人飛行載具(Unmanned Aerial Vehicles ; UAV)對挖子尾自然保留區進行多次的水線影像蒐集並產製正射影像(Orthoimage)，並透過上述的技術對現地的影像進行水線辨識分析，再以截取獲得的水線位置結合潮汐水位以內插方式獲得全域的數值高程模型(DEM)。研究結果同樣表明正射影像強化處理能夠提升OBIA得到的水線辨識精度，且整體上OBIA擁有最佳的水線辨識精度。此外，透過上述分析流程得到的高程資料能夠取得良好的精度水準，且能夠有效地測量地勢較低區域的高程值，驗證了本研究提出的方法及流程的可靠性。本研究結果證明結合水線位置與水位資料產製的數值高程模型具備快速、有效且低成本的特性，並有潛力追溯建立歷史地形，對於海岸濕地的監測管理上預期可提供正面的幫助。

關鍵字

海岸濕地 ； 影像分類 ； 水線 ； 遙測影像 ； 物件式導向分析技術

並列摘要

Wetlands are one of the most productive ecosystems. Among them, coastal wetlands are precious blue carbon ecosystems, including mangroves, seagrass beds, coral reefs. However, with the excessive coastal development in recent years, which causing coastal wetlands loss and degradation. The immediate actions for monitoring and management are necessary of coastal wetlands. Obtaining wetland elevation of coastal wetlands for establishing a digital elevation model (DEM) is a central issue to the management of natural resources. However, the current topographic measurement technology requires a large amount of manpower and funds, so as to be difficult to obtain and retrieve topographic data quickly and repeatedly. The key technology to trace and establish historical DEM through aerial imagery is scarce as well. This study proposes a novel methodology of quickly constructing a coastal wetland DEM with the instantaneous waterline combined with the tidal water level records. Correctly identify the spatiotemporal changes of waterline is a critical challenge of this research. In order to examine the waterline identification process, this study established a pilot experiment through the application of tracer dye and image enhancement to the overall classification results. This study compares the difference between Pixel-based Classification and Object-Based Image Analysis (OBIA) in the interpretation of images. This research also use noise filtering and Mathematical Morphology methods to improve the accuracy of the position of the intercepted waterline. The Edge Detection technique help intercept the position of the waterline for constructing waterline DEM. The pilot experiment results found that in most cases, water dyeing can improve the overall waterline identification accuracy, and image enhancement can effectively improve the accuracy of waterline location identified by OBIA. In addition, this research also uses Unmanned Aerial Vehicles (UAV) to collect multiple instantaneous waterline images of the Wazihwei Natural Reserve. The technique mentioned above analyzes the on-site waterline interpretation analysis of the images. And then uses the intercepted waterline position combined with the tidewater level to obtain a DEM map by interpolation. The results also show that image enhancement can improve the accuracy of waterline location identified by OBIA in most cases, and OBIA has the best waterline identification accuracy. Our results proved that the method proposed by this study offers fast, effective, low-cost aerial images identification and has the capability to establish historical topography, which is helpful for coastal wetlands in integrated management.

並列關鍵字

coastal wetlands ； image classification ； remote sensing images ； waterline ； OBIA

參考文獻

Akturk, E., Altunel, A. O. (2019). Accuracy assessment of a low-cost UAV derived digital elevation model (DEM) in a highly broken and vegetated terrain. Measurement, 136, 382-386.

Blaschke, T., Lang, S., Lorup, E., Strobl, J., Zeil, P. (2000). Object-oriented image processing in an integrated GIS/remote sensing environment and perspectives for environmental applications. Environmental information for planning, politics and the public, 2, 555-570.

Bovik, A. C., Huang, T. S., Munson, D. C. (1987). The effect of median filtering on edge estimation and detection. IEEE Transactions on pattern analysis and machine intelligence(2), 181-194.

Camp, L., Camp, R. R. (1970). Underwater acoustics. John Wiley Sons.

Chaovalit, P., Zhou, L. (2005). Movie review mining: A comparison between supervised and unsupervised classification approaches. Proceedings of the 38th annual Hawaii international conference on system sciences,