摘 要
衛星影像具有大範圍、重複觀測等優點，因此愈來愈多資源監測工作逐漸以衛星影像來作業。資源監測工作中，判釋與顯圖是很重要的，而目前使用範圍甚廣的SPOT衛星在色彩上卻只能呈現假色的影像，無法呈現出人眼所熟悉的自然色影像。本研究目的是將SPOT假色影像轉換為自然色的影像。
本研究使用倒傳遞神經網路的演算法使SPOT假色影像轉換到SPOT自然色影像。研究中分為三個測試階段，分別是（1）以同一天的Landsat假色影像與Landsat自然色影像進行訓練，得到一組權係數。（2）以該組網路測試與Landsat同一天的SPOT假色影像產生SPOT自然色影像的結果。（3）以不同天的SPOT假色影像進行網路回想，產生SPOT自然色影像。
以上三組的結果在目視上所呈現的色調皆與Landsat自然色很近似，若以相似性的評估檢核成果，相關係數皆大於0.90，顯示以類神經網路產生自然色影像確實可行，於提高SPOT影像使用層面上，能有實質成效。

Abstract
Satellite images are advantageous to large-area and repetitive observation, and therefore, are gradually adopted in the tasks of resource monitoring. Their analysis and presentation for the present are important to resource monitoring, but the most widely used SPOT satellite images can only show the image of false color, not the true color that is familiar to human eyes. This study attempts to develop a back propagation neural network method to transform a SPOT false color image into a SPOT simulate true color image.
This study consists of three steps to test the neural network method：（1）in the same period, use the Landsat false color image and the Landsat true color image in the training of neural network to get a weighting,（2）put the SPOT false image into the neural network of input layer, and create a simulated SPOT true color image,（3）use the SPOT false image of different periods with Landsat image to test the result.
The above results indicate that by visualization and mathematical testing, the presented colors are similar to Landsat nature color and their correlation coefficients are greater than 0.90. It means that this experiment is workable when we try to use simulated neural network to produce true color images. The application of SPOT true color image is certainty efficient.

參 考 文 獻
葉怡成，「類神經網路模式應用與實作」，儒林出版社，(2000)。
蔡宏宜，”SPOT衛星自然色模擬影像自動化產生之研究”，碩士論
文，國立中央大學土木工程研究所，（1998）。
Erdas , Inc . “Erdas Field Guide Fifth Edition Revised and Expanded ,” 1999.
Fang, Q. ,”Opening the Black Box of Neural Networks with Fuzzy Set Theory to Facilitate the Understanding of Remote Sensing Image Processing”,IEEE,pp.1531-1533,2000.
Graupe, D., “Principles of Artificial Neural Networks”, 1997.
Griffiths, G. H.,”Monitoring Urban Change from Landsat TM and SPOT Satellite Imagery by Image Differencing”,Proceedings of IGARSS ’88 Symposium,Edinburgh,Scotland,13-16 Sept.1988.
Hassoun, M. H.,”Fundamentals of Artificial Neural Networks “, p.57-65 . 1994.
Lillesand, T. M. , R. W. Kiefer, “Remote Sening and Image Interpretation”
, 1994.
Moik, J. G., “Digital Processing of Remotely Sensed Images” , 1980.
Richards, J. A. , “Remote Sensing Digital Image Analysis”, 1986.
Rolland, J. P. , V. Vo , B. Bloss , C. K. Abbey, “ Fast Algorithms for Histogram Matching :Application to Texture Synthesis “ ,Journal of Electronic Imaging,vol. 9,pp. 39-45,January 2000.
Zhou, W. Y.,”Verification of the Nonparametric Characteristics of Backpropagation Neural Networks for Image Classification”, IEEE Transactions on Geoscince and Remote Sensing, Vol. 37, No. 2, March 1999 .