稻株含氮量多光譜影像遙測系統之研究

Abstract

本研究以多光譜影像遙測技術為基礎，自行開發出稻株含氮量多光譜影像遙測系統。在硬體研製整合上，稻株含氮量多光譜影像遙測系統成功結合多光譜影像擷取系統、定位系統及多功能田問載具系統。在軟體開發上，利用Mierosoft VisualC++ 6.0進行程式撰寫，並與Matlab 6.0及ArcView 3.1 整合，功能包含多光譜影像之擷取、定位訊號之擷取、定位訊號之座標轉換、地理資訊系統、多光譜影像分析、迴歸分析及等高線繪製。本研究採用即時動態衛星定位系統（RTK-GPS ) ，靜態定位精度分析指出x 方向（東西）最大誤差為2.5177公分，Y方向（南北）最大誤差為2.3573 公分，顯示RTK-GPS具有極準確的定位精度，利用此遙測系統提供量測平台，以便建立之水稻含氮量多光譜影像檢量模式，對於92年一期作之台農67號水稻，採用555、680及780nm三個波長進行檢量模式分析，可得到rc=0.7667 、SEC= 0.1958、rp =0.7085、SEP=0.2252之最佳結果。對於92 年一期作之台梗14 號水稻，採用555、680 及780nm三個波長進行檢量模式分析，可得到rc=0.8032、SEC =0.1441、rp＝0.7285 、SEP=0.1734之最佳結果。本研究所開發之稻株含氮量多光譜影像遙測系統，能夠以非接觸、非破壞方式擷取水稻之多光譜影像，並藉由自行開發之軟體進行分析，最後輸出水稻田區之含氮量分佈圖，提供施肥的依據，為精準農業提供一套有效且準確的遙測系統。 This study is aimed to develop a multi-spectral remote imaging system to measure the nitrogen content in rice. In the integration of the hardware, the multi-spectral imaging system, the position system and the multi-function field-carrier were successfully integrated to develop a ground-based remote sensing system. As for the software, Microsoft Visual C++ 6.0, Matlab 6.0 and ArcView 3.1 were used to develop a program. The program was implemented with functions including acquisition of the multi-spectral images, collection of the position signals, transformation of geographic information, processing of multi-spectral images, analysis of model regressions and expression of contours. The study used the RTK-GPS (Real-Time Kinematic GPS) as position system, with an accuracy of 2.52 and 2.36 cm in X- and Y-direction respectively. The multi-spectral imaging model is developed by using the multi-spectral remote imaging system as a measurement platform. Regarding rice of Tainung 67 variety (ist crop, 2003), the model used three band wavelengths (555, 680 and 780 nm) to predict the nitrogen content in rice. The model gave the results as r = 0.7667, SEC = 0.1958, r = 0.7085 and SEP = 0.2252. As to rice of Taiken 14(1st crop, 2003), the model also using wavelengths at 555, 680 and 780 nm showed the results of r = 0.8032, SEC = 0.1441, r = 0.7285 and SEP = 0.1734. The study developed a multi-spectral remote imaging system to measure nitrogen content in rice, which is capable to obtain the multi-spectral images in non-contact and non-destruction way, and the nitrogen content distribution can be analyzed in rice fields by the using the self-developed program. The system will provide a basis of future application to site-specific management of fertilizer, and to the precision agriculture.