摘 要
本研究中主要使用雷達影像來進行變遷偵測，利用合成孔徑雷達衛星的全天候不受天氣影響及多偏極的特性，在每個偏極方向都包含了資訊，為了將全部偏極的資訊結合起來，便將影像做主軸的轉換。傳統主軸轉換是使用差值影像的共變異矩陣所進行的主成分分析。本研究以典型相關分析(Canonical Correlations Analysis)為基礎，反演為多變數轉化偵測法(Multivariate Alteration Detection)，而此方法的特色在於此轉換方法具有線性轉換的不變性，即使兩時期影像因時間的不同有不同的背景輻射強度，或因儀器本身的補償(offsets)與增益(gain)這些線性關係的影響，也不需要作絕對的校正前處理。再以轉換後影像，利用卡方統計檢定法，判斷變遷區域，使變遷偵測的結果能更加接近實際的變遷。
研究中利用多變數轉化偵測法(Multivariate Alteration Detection)，來觀察以人為改變後期影像部分區塊後的模擬結果，最後再利用ENVISAT及STOP兩組衛星影像來做變遷測試，而測試結果可以利用典型變數來觀察與原始影像每個波段或每個偏極方向的關係。

Abstract
SAR image is becoming more important data source for change detection, because of its advantages of all-weather、day-and-night operations and providing multi-polarization information. To perform change detection, linear transformations of the image data is usually adopted, for example, most commonly, by principal component analysis. In this study, we used the multivariate alteration detection (MAD) to perform the transformation of the two image sets. One of advantages of the multivariate alteration detection (MAD) transformation is its the linear scale invariance. This means that it is not sensitive to the offsets or gain settings of a measuring device, or to radiometric and atmospheric correction schemes that show a linear relationship with brightness counts. The change, after MAD transformation, was carried out by Chi-Square test. To verify and validate the procedure, we first used simulation images at different time and artificially making land cover changes at different polarizations. Finally, the method was test on real images from SPOT and ENVISAT multi-polarization data. The relationship between each band or polarization was investigated from canonical variates. The detection accuracy was found to be satisfactory.

[1] Cooley, W. W., and Lohnes, P. R.,“Multivariate Data Analysis,”Wiley, New York, 1971
[2] Anderson, T. W., “An Introduction to Multivariate Statistical Analysis, 2nd ed.,” Wiley, New York, 1984
[3] Nielsen, A. A., K. Conradsen and J. J. Simpson, “Multivariate Alteration Detection (MAD) and MAF Postprocessing in Multispectral, Bitemporal Image Data: New Approaches to Change
Detection Studies,” Remote Sens. Environ., vol. 64, pp. 1-19, 1998
[4] Nielsen, A. A. , “Multiset Canonical Correlations Analysis and Multispectral,Truly Multitemporal Remote Sensing Data,” IEEE Transactions on Image Processing, 11(3), 293-305, 2002
[5] Stauffer, M.L. and R.L. McKinney, “Landsat Image Differencing asAn Automated Land Cover Change Detection Technique,” Computer Sciences Corporation, Technical Memorandum CSC/TM-78/6215
Silver Spring, MD, 1978
[6] Singh, A., “Change Detection in the Tropical Forest Environmental ofNorthern India using Landsat,” Remote Sens. and Tropical Land
Management, M.J. Eden and J.T. Parry, Eds. John Wiley & Sons,London, pp.237-254, 1986
[7] Wilson, J. R., C. Blackman, and G. W. Spann, “Land use Change Detection using Ladsat Data,” Proceedings of the 5th AnnualRemote Sensing of Earth Resources Conference,University of Tennesses, Tullhama,TN, 1976, pp.79-91
[8] Rubec, C.D., and J. Thie., “Land use Monitoring with Landsat Digital Data in Southwestern Manitoba”, Proceedings of the fifth Canadian Symposium on Remote Sensing, Victoria, BC, pp. 136-150, 1987
[9] Byrne, G.F., P.F. Crapper, and K.K. Mayo, “Monitoring Land-cover Change by Principal Component Analysis of Multitemporal Landsat
Data,” Remote Sensing Environ., vol.10, pp. 175-184, 1980
[10] Fung, T. and LeDrew, E. “Application of principal components analysis to change detection, ” Photogramm. Eng. Remote Sens,. vol.
53, pp. 1649-1658, 1987
[11] Canty, M. J., A. A. Nielsen and M. Schmidt, “Automatic Radiometric Normalization of Multitemporal Satellite Imagery,” Remote Sens. Environ., vol. 91, pp. 441-451, 2004
[12] Lawrence L.Lapin“Probability and Statistics for Modern engineering, 2nded.,”Belmont, California, 1990
[13] Du, Qian, “Noise Estimation for Remote Sensing Image Data Analysis,” SPIE, Bellingham, WA., 2003
[14] Jong-Sen Lee, Senior Member, IEEE, Mitchell R .Grunes, and Stephen A.Mango ”Speckle Reduction in Multipolarization Multifrequency SAR Imagery" IEEE Transactions on Geoscience And Remote Sensing, vol. 29, no. 4, July 1991
[15] Tzeng, Y. C., K. S. Chen, W. L. Kao and A. K. Fung, “A Dynamic Learning Neural Network for Remote Sensing Applications,”IEEE Trans. Geosci. Remote Sensing, vol. 32, no. 5, Sept. 1994
[16] Y. C. Tzeng, K. S. Chen, ”A Fuzzy neural network to SAR Image Classification” IEEE Transactions on Geoscience And Remote Sensing, vol. 36, no. 1, January 1998
[17] A. A. Nielsen, “Analysis of regularly and irregularly sampled spatial, multivariate and multi-temporal data,” Ph.D. dissertation, Dept. Inform. Math. Modeling, Tech. Univ. Denmark, Lyngby, Denmark, http://www.imn.dtu.dk/~aa/phd/, 1994.
[18] J. S. Lee, K. W. Hoppel, and S. A. Mango, “Intensity and Phase Statistics of Multilook Polarimetric and Interferometric SAR Imagery”,IEEE Trans. on Geoscience and Remote Sensing, vol.32, no.5, pp.1017-1028, September, 1998
[19] Cloude S.R., Pottier E.,“Polarimetric SAR Interferomerty”, IEEE Trans. on Geoscience and Remote Sensing, vol. 36, no.5, pp.68-78, Jan. 1997
[20] 洪志賢, “多變數轉化偵測法(MAD)應用於多光譜影像變遷偵測之研究”, 國立中央大學太空科學研究所碩士論文, 2005