臺灣博碩士論文加值系統

資源衛星影像的應用，近年來非常廣泛。但是目前使用最普遍的SPOT衛星
影像因只有綠色、紅色和近紅外波段所以只能提供假色合成的影像。在人
工判識和影像展示上，與一般人習慣的自然色影像有很大的差異，往往造
成許多不便及誤判，本研究嘗試將SPOT假色合成影像轉換至自然色波段的
影像。本研究主要是利用光譜轉換的原理，配合Fuzzy理論上的歸屬值概
念進行光譜座標的轉換。在本實驗中，應用類似傳統空間座標轉換的方法
，先選取光譜上的控制點，計算出光譜轉換係數，然後再進行光譜轉換。
共有三種不同的影像進行本研究的測試，分別是(1)Landsat衛星影像本身
的（G,R,IR）波段轉換到（B,G,R）波段來測試本研究方法的可行性，(2)
SPOT（G,R,IR）衛星影像轉換到同一天的Landsat（B,G,R）影像以及(3)
不同時間的SPOT衛星影像（地面覆蓋物有變）轉換到Landsat（B,G,R）波
段影像。測試結果顯示，本研究可將SPOT影像轉換至類似Landsat TM的自
然色影像。

Landsat TM and SPOT multispectral satellite images have been
widely used in various applications. The spectral and spatial
comparison of Landsat TM and SPOT shows that Landsat TM has a
broader spectral coverage than SPOT, especially Landsat TM has
covered a more complete visible range (blue, green, and red
bands) than SPOT (green and red bands), however, SPOT has a
higher ground resolution (20 m) than Landsat TM (30 m). Both
satellite images have been commonly used to develop the digital
techniques for image information extraction. However, the
demand of a higher resolution satellite natural color image has
significantly increased recently due to the increasing needs of
visual interpretation and display. It appears that the best
candidate currently would be the SPOT multispectral image.
However, the spectral coverage of SPOT can only produce the
false color image. This study attempts to develop a spectral
coordinate transformation method to transform a SPOT false color
image into a simulated natural color image. The method adopts
the basic concepts of the geometric transformation that consists
of four steps: (1) the selection of spectral control points, (2)
the choosing of a transformation function, (3) the finding of
the transformation coefficients, and (4) the transformation of
the spectral bands of the false image into the natural image.
The methods have been tested using Landsat TM and SPOT images.
The qualitative and quantitative comparisons indicate that the
striking similarity can be found between SPOT simulated natural
color images and Landsat TM natural color images in terms of the
visual looks and the correlation coefficients.