高解析度衛星影像之幾何特性分析

Abstract

由於衛星影像能夠於高空詳實記錄大 範圍地表之情況，因此長久以來常被用作 軍事目的、地球資源之監控與土地利用情 形之觀察或是氣象之預測。而隨著太空科 技的進步，衛星影像之幾何解析度與輻射 解析度逐漸提高，使其用途更加多元化； 尤其是於 1999 年 9 月 24 日第一顆高解析 度之商業衛星 IKONOS 成功升空並開始運 轉後，更是宣告高解析度遙測時代已然來 臨。 正當 IKONOS 衛星之高解析度影像引 起眾人注目並投入研究之際，另一顆高解 析度之商業衛星 QuickBird 亦於 2001 年 10 月 19 日成功發射，相較於 IKONOS 衛星影 像，其所拍攝之影像除了具有同樣的輻射 解析度，包括全色態（ Panchromatic）掃描 系統與多光譜掃描（ Multispectral Scanner） 系統之外，其全色態影像與多光譜影像之 幾何解析度分別為 0.7 公尺與 2.8 公尺，甚 至更高於 IKONOS 衛星影像，因此，應用 QuickBird 衛星影像於高精度之測量製圖 方面更顯見潛力。 在實際應用衛星影像之前，須先探討 與了解衛星影像的幾何特性與輻射特性。 輻射性分析是探討大氣輻射及地表反射率 與多光譜影像輻射值之關係，而幾何特性 則是探討像空間與物空間之點位對應關 係，以進行影像之幾何糾正，將衛星影像 的幾何變形減低到最小程度。據此，本研 究 即選擇 QuickBird 衛星影像為研究對 象，進行分析其幾何特性。首先分別採用 遙測影像處理軟體 PCI Geomatics 以及 Z/I Imaging，利用行政院農林航空測量所拍攝 含括實驗區之航空像片，以自動匹配的方 式製作數值表面模型 (DSM)，接著將此數 值表面模型應用於 QuickBird 衛星影像之 正射糾正，檢視其正射成果並探討其精密 幾何糾正與運用於製圖之可行性，以期將 來能利用高解析度衛星影像於製圖，取代 傳統的航測方式。

Since ERTS-1 was launched on July 23, 1972, Satellite images had been used in environmental applications. Because of the convenience and flexibility, spaceborne sensors, such as SPOT or Landsat, are broadly used to assist in tasks including weather prediction, crop forecasting, mineral exploration, pollution detection, rangeland monitoring, etc. However, the ground resolution of images taken by most of the commercial satellites is not high enough to be used for mapping until IKONOS was launched. IKONOS satellite was launched in 1999 and it can take data in panchromatic mode with 1m or in multi-spectral mode with 4m. Because improving in ground resolution, IKONOS images may be applied in map-making. QuickBird satellite was launched in 2001, could also take data with higher ground resolution than IKONOS (panchromatic resolution is 0.7m, multispectral is 2.8m) did. Due to its high ground resolution, aerial photos for mapping could be replaced by QuickBird satellite images. In this research, the QuickBird satellite images were first ortho-rectified, using the digital surface models (DSM) which were extracted automatically by Z/I Imaging and PCI Geomatics software. The resultant images were then compared with the existed orthophotomap to determine the potential in applying satellite images in map-making.