多波段影像特徵分析於灘線萃取之研究

Extraction of Shorelines from Multi-Spectral Images Using Feature Analysis

陳良健(Liang-Chien Chen) ； 徐啟崇(Chii-Chorng Shyu)

《航測及遙測學刊》 4卷3期 (1999/09) Pp. 17-30

https://doi.org/10.6574/JPRS.1999.4(3).2

摘要

海岸之監測為國家發展及環境保育的重要工作。其中灘線之萃取更是最基本的工作項目。本研究目標即針對多波段衛星影像進行海岸線萃取。本研究對於海岸線之萃取以SPOT為資料來源。提出之方法其主要概念為：（1）經由影像分割的方式，先將影像區分為水陸二元圖，再經由邊界資訊萃取，來建立海岸線的近 似位置；（2）利用所建立海岸線之近似位置作為精確海岸線萃取之預估區間，在該區間內進行邊緣線之偵測及追蹤，以萃取出精確海岸線位置。將所萃取的海岸線位置與（1）人工數化之成果進行比對，及（2）實際野外GPS量測海岸 線位置做比對，以檢核萃取出海岸線位置之正確性。測試成果顯示海岸線萃取位置與人工數化成果相較，就差異之像元數而言為2.48%。與GPS量測海岸線之間的差距在1~2像元間。

關鍵字

多波段影像 ； 影像分割 ； 灘線萃取

並列摘要

Coastal zone monitoring is an important task in national development and environmental protection. In this, the extraction of shorelines is a fundamental work. This investigation aims to extract shorelines in an automated way for optical images.SPOT multi-spectral imagery is the major data source in this investigation for shoreline extraction. The main components of extraction for shorelines include: (1) obtaining a rough separation between the land and water and (2) refining the rough land-water boundaries by edge detection and edge tracing algorithms to extract the accurate shoreline position.To assess the accuracy of shoreline extracted automatically from the imagery, two validations are included: (1) the comparison of the extracted shorelines to the ones digitized by manual operation and (2) ground measurements by GPS were treated as references. The shoreline positions determined from the images were found to agree with (1) the manual digitization in 2.48% in terms of area, and (2) the ground measurements to within one to two pixels.

並列關鍵字

Multi-spectral image ； Segmentation ； Shoreline extraction