臺灣博碩士論文加值系統

由於海洋汙染對生態迫害及經濟損失的衝擊甚大，快速且準確的找出海洋污染區域是迫切需要的。隨著遙測技術的進步，衛星影像已被廣泛地使用於偵測海洋污染(異常物)。而異常物與背景物的反射率有顯著的不同，本研究的目的就是利用此特性進行自動化海上異常物偵測(anomaly detection)。
本研究方法，分成三大步驟：(1) 首先利用 RX ( Reed & Xiaoli )演算法產生異常物強度影像，影像中強度值越高的像元，屬於異常物的機率越高；(2) 將異常物強度影像利用期望值最大化演算法 (Expectation-Maximization, EM)，求得異常物和背景兩類各自的機率密度函數，基於貝氏理論 (Bayes Rule) 在兩曲線相交的地方，兩類別誤差合為最小，則為最佳門檻值，高於門檻值部分視為異常物，反之則為背景，進而產生初始二值化影像；(3) 選定誤授率相對較低的像素的異常物點作為種子點，利用區域成長法( Region Growing )在初始二值化影像上，找出異常物主體部分，得到成果影像。
本研究使用SPOT-4、SPOT-5 及福衛二號光學衛星影像作為測試資料，和傳統給定門檻值方法及人工數化成果比較後，顯示本研究方法可以自動化且完整偵測出海上異常物，同時能快速的得到異常物主體形狀、面積以及坐標，利於使用者決策及解決問題，免去人工數化之繁瑣過程。

Since the ocean pollution usually causes severe damage to the environment and economy, it is the important issue for detecting the pollution rapidly. The satellite imagery could provide observations of wide areas; it can be used for detecting the anomaly which may be the pollution on the ocean. Generally, the anomaly is defined as the object with different characteristics of reflectance from the background. Base on this definition, the automatic algorithm could be developed to detect the anomaly.
In this study, a three-stage algorithm is proposed to detect the anomaly automatically: (1) Use RX (Reed & Xiaoli) algorithm to derive the RX image which represents the intensity of the anomaly. (2) Use EM (Expectation-Maximization) algorithm to classify the image into two probability distribution functions which represents the anomaly and background respectively. Then a threshold is determined to binarize the image to show up the anomaly. (3) In order to reduce the noise, the region growing algorithm is used to refine the anomaly image.
Various satellite images are used to test the proposed algorithm. The results show that the shape, area, and location of ocean anomaly could be observed clearly and accurately. Furthermore, the accuracy information could be estimated to evaluate the result.

摘要 i
Abstract ii
誌謝 iii
目錄 iv
圖目錄 vii
表目錄 xi
第一章 前言 1
1-1 研究背景與目的 1
1-2 文獻回顧 3
1-2-1 海洋異常物偵測 3
1-2-2 RX 演算法 4
1-2-3 門檻值選定 5
1-2-4 期望值最大化演算法( Expectation-Maximization, EM ) 7
1-2-5 區域成長法( Region Growing ) 8
1-3 研究內容與論文架構 9
第二章 研究方法 10
2-1 RX 演算法 11
2-2 門檻值之選定 13
2-3 期望值最大化演算法(Expectation-Maximization) 14
2-4 空間過濾濾除雜訊 17
2-4-1 區域成長法( Region Growing ) 18
2-5 成果分析 20
第三章 測試資料介紹 23
3-1 測試影像介紹 23
第四章 研究成果與分析 32
4-1 本研究成果 32
4-1-1 RX 演算法成果 32
4-1-2 EM 演算法成果 38
4-1-3 利用空間過濾消除雜訊成果 49
4-2 傳統方法與本研究方法成果比較 62
4-3 人工數化成果與本研究方法成果比較 73
第五章 結論與建議 81
5-1 結論 81
5-2 建議 82
參考文獻 84

吳基、莊甲子，1996，“利用SAR 影像於海面油污偵測”，第二次資源衛星資料應用研討會。
林明顯，1997，“應用基波轉換於SAR 影像之油污邊緣偵測”，國立台灣海洋大學碩士論文。
許君韶，2005，“區塊分割變遷偵測法於多時期衛星影像之應用”，國立中央大學碩士論文。
張智安、陳良健，2006，“利用光達資料模塑建物之研究”，航測及遙測學刊11卷2期。
劉書彰，2005，“臺灣附近海域油污染之衛星觀測”，國立台灣海洋大學碩士論文。
Brekke, C. and A. H. S. Solberg, 2005. “Oil spill detection by satellite remote sensing,” Remote Sensing of Environment, Vol.95, pp. 1-13.
Chang, C. –I, S. S. Chiang, I. W. Ginsberg, 2001. “Anomaly Detection in Hyperspectral Imagery,” Proceedings of SPIE, Vol.4383, pp. 43-50.
Dempster, A. P., N. M. Laird, and D. B. Rubin, 1977. « Maximum Likelihood from incomplete data via the EM algorithm,” Journal of the Royal Statistical Society, Series B, Vol. 39, pp.1-38.
Fingas, M. F. and C. E. Brown, 1997. “Review of oil spill remote sensing,” Spill Science & Technology Bulletin, Vol. 4, pp. 199-208.
Gade, M. and W. Alpers, 1999. “Using ERS-2 SAR images for routine observation of marine pollution in European coastal waters,” The Science of the Total Environment, 237/238, pp. 441- 448.
Gonzalez, R. C. and R. E. Woods, 2002. “Digital Image Processing,” 2nd ed.: Prentice Hall, Inc.
Fukunaga, K., 1990. “Introduction to Statistical Pattern Recognition,” 2nd ed.: San Diego, 591 p.
Justice, R. and E. Stokely, 1996. “3-D Segmentation of MR Brain Images Using Seeded Region Growing,” 18th Annual International Conference of the IEEE Engineering in Medicine and Biology Society , pp. 1083-1084.
Kutser, T., A.G. Dekker, W. Skirving, 2003. “Modeling spectral discrimination of GreatBarrier Reef benthic communities by remote sensing instruments,” Limnologyand Oceanography Vol.48, pp. 497-510.
Kwon, S. H., 2004. “Threshold selection based on cluster analysis,” Pattern Recognition Letters, v.25 n.9, p.1045-1050.
Lloyd, S. P., 1982. “Least squares quantization in PCM,” IEEE Trans. Inform. Theory, vol. IT-28, pp. 129–137.
Lievin, M., N. Hanssen, P. Zerfass, 2001. “3D Markov Random Fields and Region Growing for Interactive Segmentation of MR Data,” 4th International Conference on Medical Image Computing and Computer-Assisted Intervention , pp. 14-17.
McLachlan, G. J., and T. Krishnan, 1997. “The EM algorithm and extensions,” New York: John Wiley & Sons.
Mumby, P. J., W. Skirving, A. E. Strong, J. T. Hardy, E. F. LeDrew, E. J. Hochberg, R. P. Stumpf, L.T. David ,2004. “Remote sensing of coral reefs and their physical environment,” Marine Pollution Bulletin Vol.48, pp. 219-228.
Otsu, N., 1979. “A threshold selection method from gray-level histograms,” IEEE Trans. Sys., Man., Cyber., vol. 9, pp. 62–66.
Reed, I. and X. Yu, 1990. “Adaptive Multiple-Band CFAR Detection of an Optical Pattern with Unknown Spectral Distribution,” IEEE Trans. on Acoustics, Speech and Signal Process, vol.38, no.10, pp. 1760-1770.
Ren, H., Q. Du, and J. Jensen, 2002. “Efficient anomaly detection and discrimination for hyperspectral imagery,＂Proc. SPIE, vol.4725, pp. 234-241.
Riegl, B., 2004. “Remote sensing: a key tool for interdisciplinary assessment of coral reef processes,” Coral Reefs Vol23, pp. 1- 4.
Srinath, M. D., P. K. Rajasekaran, R. Viswanathan, 1996. “Introduction to statistical signal processing with applications,” Prentice-Hall, Inc., pp.146-149.
Tao, Y., W. Grosky, L. Zamorano, 1999. “Segmentation and Representation of Lesions in the MRI Brain Images”, Proccedings of SPIE Medical Imaging , pp. 930-939.