完整後設資料紀錄
DC 欄位語言
dc.contributor.author侯尚儒en_US
dc.contributor.authorHou, Shang-Juen_US
dc.contributor.author史天元en_US
dc.contributor.authorShih, Tian-Yuanen_US
dc.date.accessioned2014-12-12T02:39:45Z-
dc.date.available2014-12-12T02:39:45Z-
dc.date.issued2013en_US
dc.identifier.urihttp://140.113.39.130/cdrfb3/record/nctu/#GT070051283en_US
dc.identifier.urihttp://hdl.handle.net/11536/74081-
dc.description.abstract珊瑚礁是由珊瑚不斷生長、堆積而成,依其地形構造,可以分成礁原和礁斜面兩大部分,依其形態則可分成裙礁、堡礁、及環礁。本研究之區域為東沙環礁,目的是偵測並繪製東沙環礁內礁塊的分布。 本研究所使用的數據資料,包含數值高程模型(DEM)與衛星影像,研究區域依地貌特徵選取三幅 1/5,000圖幅,其位置分別為潟湖區中央、潟湖區與礁台上的交界處、與礁台上。 本研究採用之偵測礁塊位置,包含地形趨勢面分析與物件導向分割,前者僅能以DEM進行偵測,而後者則可應用於DEM與衛星影像及兩者整合。以自動方式偵測出的結果,與以人工目視DEM判斷、手繪出的塊礁分布比對。在各種方法與組合中,以結合DEM與影像進行物件導向分割的成果最佳。其中五組不同比重的分析中,又以DEM比重為5,影像比重為1的效果最佳,三個研究區域的平均正確率高達0.97以上,Kappa值為0.84以上。zh_TW
dc.description.abstractCoral reef is formed by the corals’ growth and accumulation. According to the geography, it can be divided into two parts: the flat and the slope. By its construction, coral reef can be divided into three types: the fringing reef, the barrier reef and the atoll reef. The atoll reef areas on Dongsha Island are what being researched in this study, and this research aims at detecting and drawing the location of reefs around the Dongsha atoll area. The data used in this research includes digital elevation model (DEM) and satellite images. Three 1/5000 drawings had been chosen as the researched area depending on the characteristics of terrain, and their locations are: the middle of the lagoon area, the junction of the lagoon and the reef plat, and the reef plat. Trend surface analysis and object based segmentation are the two reef location-detecting ways that this research uses. The first one can only detect the location of reefs in DEM, while the second can detect both in DEM and the satellite images. The detected results will be compared with the real situation basing on the human observation of DEM and the sketched map of the reefs distribution. According to the different ways and combinations, it proves that object based segmentation in combination of the image of DEM, serves best. Among the five different analyses, the best one is DEM weight of 4 and image weight of 1. The average accuracy of the three areas is above 0.97, and the Kappa coefficient is higher than 0.84.en_US
dc.language.isozh_TWen_US
dc.subject趨勢面zh_TW
dc.subject分割zh_TW
dc.subject珊瑚礁zh_TW
dc.subjecttrend surfaceen_US
dc.subjectsegmentationen_US
dc.subjectcoral reefen_US
dc.title珊瑚環礁塊礁分佈圖製作研究zh_TW
dc.titleCoral Reef Patch Mappingen_US
dc.typeThesisen_US
dc.contributor.department土木工程系所zh_TW
顯示於類別:畢業論文