2016-01-012024-05-17https://scholars.lib.ntu.edu.tw/handle/123456789/684281摘要：自民國80年起，國內部分學者開始利用衛星影像及GIS資料，發展自動化水稻種植範圍影像判釋，判釋正確性已從早期的60%左右，逐步提升至96%以上。但在航遙測資料持續發展的環境下，影像資料已不僅只有光學衛星與航照相機，可見光/多光譜無人飛行載具與雷達衛星影像，甚至無人飛行載具雷達影像，都可更進一步提供多時段、多光譜與多高程航遙測資料中的水稻生長資訊。現今國內外的學者便藉由如此便利又多元的航遙測資料，以不同的演算法與資料組合，發展出多種農作物的影像判釋與產量預估模式。 國內現行的水稻種植分佈與面積調查方式，主要仍以農航所數值航照人工判釋，再輔以地面抽樣驗證判釋正確性。由於農航所自民國97年已利用數位製圖像機及空載數位掃描儀，同步紀錄4波段的地表光譜資訊，因此已具備發展自動化農作物航照影像判釋的基礎。本研究將以歷年利用光學衛星影像進行水稻種植範圍自動判釋經驗 與成果為基礎，結合103年發展之以L-Band雷達影像進行水稻判釋模式、102-104年所發展之多元多期空間資訊整合分析模式，及前人以航遙測資料進行稻作估產研究成果，研發以多元航遙測資料進行水稻種植面積自動判釋與產量推估模式。 藉由本研究所建立的水稻種植面積自動判釋與產量推估模式，將國內可有效取得與運用的航遙測資料，包含光學衛星、4波段數值航照、可見光/近紅外光無人飛行載具、雷達衛星影像資料，甚至即將引進國內的無人飛行載具雷達影像，整合後進行自動化且高精度的水稻種植面積判釋與產量推估，藉以提升國內的水稻種植現況調查與產量估算的整體效益。 <br> Abstract: From 1990s, some researchers were used satellite images and GIS data to develop automatic paddy rice interpretation in Taiewan. The interpret accuracy was started from 60%, increased to 96% up via model improvement. With remote sensing equirements developing, besides optical satellite image and aerial photo, we can apply UAV VIS/NIR images, satellite SAR images and even UAV SAR images, to analyze growing information about paddy rice by multi-temporal, multi-spectral and multi-evelation remote sensing data. With these data and different algorithms, researches can develop image interpretation and yield prediction models for several crops. In Taiwan, paddy rice investigation are mainly accomplished by manually interpretion with aerial photo from Aerial Survey Office (ASO) and sampling verification with ground survey. From 2008, ASO started to use Digital Mapping Camera (DMC) and Aerial Digital Sensor (ADS) to record 4 bands (Red, Green, Blue, NIR) information of land cover in Taiwan. So that, it is ready to develop crop interpretation with digital aerial images from ASO. In this research, with the experience of automatic interpretation paddy rice from optical satellite images in recent years, L-Band SAR image interpret method for paddy rice, multi-sources and multi-temporal spatial information integrated analysis method and the research results of rice yield prediction with remote sensing data, we will develop a method of automatic paddy rice interpretation and yeild prediction from multi-sources remote sensing data. Applying this method, we can efficientally use apporpriate remote sensing data, such as optical satellite/aerial image, satellite SAR image and UAV VIS/NIR/SAR images, and can integrate into automatic and high accuarcy paddy rice interpretation and yeild prediction method to improve the investigation and yeild prediction benifit of paddy rice in Taiwan.多元航遙測資料水稻產量推估分類誤差分析影像特徵萃取Multi-Sources Remote Sensing DataRice Yield PredictionClassification Error AnalysisImage Feature Extraction