現在位置首頁 > 博碩士論文 > 詳目
論文中文名稱:以基於互訊息的二維非純度波段優先權方法評估高光譜影像波段選取在不同分佈下的影響 [以論文名稱查詢館藏系統]
論文英文名稱:Evaluation of 2-Dimension Impurity Function Band Prioritization Using Mutual Information for Hyperspectral Images under Different Distributions [以論文名稱查詢館藏系統]
院校名稱:臺北科技大學
學院名稱:電資學院
系所名稱:電機工程系
畢業學年度:106
畢業學期:第二學期
出版年度:107
中文姓名:潘俊桂
英文姓名:Chun-Kuei Pan
研究生學號:103318100
學位類別:碩士
語文別:中文
口試日期:2016/07/27
論文頁數:44
指導教授中文名:張陽郎;方志鵬
口試委員中文名:王怡鈞;鞠志遠
中文關鍵詞:高光譜影像常態分佈均勻分佈指數分佈T分佈降維波段選取非純度維度優先權法互訊息
英文關鍵詞:hyperspectral imagesnormal distributionuniform distributionexponential distributionT distributiondimension reductionband selectionimpurity function band prioritization (IFBP)mutual information
論文中文摘要:在衛星遙測技術逐年進步下,使得近年來衛星遙測影像的波段跟資料量不斷地增加,為解決高光譜影像之波段與資料量過於龐大的問題,可利用波段選取的方式來降低影像的維度,避免波段數量過多導致Hughes現象發生,以及減少大量的運算時間。
以往有學者提出二維的非純度維度優先權法(2D - Impurity Function Band Prioritization, IFBP),由二維平面上來看類別之間的覆蓋率,並運用互訊息(Mutual Information, MI)之中的機率概念,套入二維常態分佈來計算兩個維度類別間的覆蓋程度,再透過計算出來的2D-IFBP去修正IFBP的波段順序,並獲得極佳的降維結果。因此本論文將架構在此成果之上,加入更多不同的二維機率分佈,包括均勻分佈、指數分佈和T分佈,並進行綜合地探討,嘗試找出不同的資料分佈在應用互訊息降維上的影響。
本論文使用Salinas、Washington DC mall以及Pavia University等三套遙測影像作為實驗的圖資,由實驗結果得知前兩套圖資都是呈現常態分佈,但是在Pavia University圖資上使用常態分佈的結果並非最為理想。取而代之的是T分佈,在與常態分佈相同降維率的情況下,其能夠擁有更高的正確率。因此若能針對不同的圖資,予以採用適當的分佈,將會比任何一套圖資都使用常態分佈來的適切。
論文英文摘要:In recent years, the technology of remote sensing has improved significantly. To solve the problem of over-sized data and bands, band selection method has been widely utilized to reduce the dimensions of images. In this way, Hughes phenomenon can be avoided and the computation time can also be reduced.
Two-dimension Impurity Function Band Prioritization (2D-IFBP) method has been proposed by other researcher. It utilizes Mutual Information (MI) and bivariate normal distribution to calculate the overlapping rate between two classes from the two-dimensional domain. In addition, 2D-IFBP is used to modify the band order of IFBP to obtain the better result. In this thesis, some other bivariate distributions will also be applied such as uniform distribution, exponential distribution, and T distribution to evaluate its characteristic of Mutual Information.
In this experiment, the data sets of Salinas, Washington DC mall and Pavia University are used to verify the influence of different distributions on 2D-IFBP. The experimental results show that the first and the second data sets are more like normal distribution but the data set of Pavia University is more like T distribution. Moreover, T distribution has the same dimension reduction rate as normal distribution but it possesses the better accuracy rate than the other one. Hence, trying to utilize different distributions for the specific data sets is more appropriate than conventional method.
論文目次:目錄

摘要 i
ABSTRACT ii
誌謝 iii
目錄 iv
表目錄 vi
圖目錄 vii
第一章 緒論 1
1.1 研究背景介紹 1
1.2 研究動機與目的 3
1.3 論文內容大綱 5
第二章 相關文獻回顧 6
2.1 高光譜影像介紹 6
2.2 相關係數與相關係數矩陣 8
2.3貪婪模組特徵空間 9
2.4 粒子群優演算法 11
2.4.2 粒子群優法流程 12
2.4.3 粒子移動公式及參數 13
2.5 適應值函數及空間轉換 15
2.6 非純度波段優先權法 17
2.7互訊息理論 18
第三章 研究方法 21
3.1 高光譜影像分類方法 21
3.2 IFBP波段選取方法 22
3.3基於互訊息之2維-非純度波段優先權 24
3.4使用分佈簡介 29
3.5參考指標 30
3.6 PSO參數設定 31
第四章 實驗結果 32
4.1 使用圖資介紹 32
4.1.1 Washington DC Mall (WDC) 32
4.1.2 Salinas 34
4.1.3 Pavia University 36
4.2 實驗目的與實驗環境 37
4.3 實驗結果 38
第五章 結論與未來發展 42
5.1 結論 42
5.2 未來展望 42
參考文獻 43
論文參考文獻:[1] 林惠玲和陳正倉,統計學-方法與應用(上冊),第314-325頁,雙葉書廊有限公司,臺北,2000。
[2] G. F. Hughes, “On the mean accuracy of statistical pattern recognizers,” IEEE Transactions on Information Theory, vol. 14, no. 1, pp. 55–63, 1968.
[3] C. Lee and D. A. Landgrebe, “Analyzing high-dimensional multispectral data,” IEEE Trans. Geosci. Remote Sensing 31, no. 4, pp. 792–800, 1993.
[4] C.-I. Chang, Q. Du, T.-L. Sun, and M. L. G. Althouse, “A Joint Band Prioritization and Band Decorrelation Approach to Band Selection for Hyperspectral Image Classification,” IEEE Trans. Geosci. and Remote Sens., vol. 37, no. 6, pp. 2631-2641, 1999.
[5] C. E. Shannon, “A mathematical theory of communication,” Bell System Technical Journal, vol. 27, no. 4, pp. 623–656, 1948.
[6] Y. L. Chang, C. C. Han, K. C. Fan, K. S. Chen, C. T. Chen, and J. H. Chang, “Greedy modular eigenspaces and positive boolean function for supervised hyperspectral image classification,” Opt. Eng. 42, no. 9, pp. 2576–2587, 2003.
[7] J. Kennedy and R. C. Eberhart, “Particle swarm optimization,” Proc. of the IEEE, Int. Conf. on Neural Networks, pp. 1942–1948, 1995.
[8] R. C. Eberhart and J. Kennedy, “A new optimizer using particle swarm theory,” Proc. IEEE Int. Symposium on Micro Machine and Human Science, Nagoya, Japan, pp. 39-43, 1995.
[9] Wei Pang, Kang-Ping Wang, Chun-Guang Zhou, “Modified Particle Swarm Optimization Based On Space Transformation For Solving Traveling Salesman Problem,” Machine Learning and Cybernetics, 2004. Proceedings of 2004 International Conference, vol. 4, pp. 2342-2346, 2004.
[10] Yang-Lang Chang, Bin-Feng Shu, Tung-Ju Hsieh, Chih-Yuan Chu and Jyh-Perng Fang, “Band selection for hyperspectral images based on impurity function,” IEEE International Geoscience and Remote Sensing Symposium (IGARSS), Vancouver, USA, pp. 2369 - 2372, 2011 .
[11] 徐斌峰,一個維度優先權方法應用於粒子群優法在高維度影像特徵抽取,碩士論文,國立台北科技大學電機工程研究所,臺北,2011。
[12] 王旭,基於GPU之粒子群優法應用於高光譜影像波段選取,碩士論文,國立台北科技大學電機工程研究所,臺北,2012。
[13] 王宏原,平行粒子群優法應用於高維度影像特徵抽取,碩士論文,國立臺北科技大學電機工程研究所,臺北,2010。
[14] 簡士傑,一個基於互訊息的二維非純度波段優先權方法應用於高光譜影像波段選取,碩士論文,國立臺北科技大學電機工程研究所,臺北,2015。
[15] 國土資訊系統通訊 http://ngis.moi.gov.tw/lan_ch/
[16] 基本光谱分析http://blog.sina.com.cn/s/blog_764b1e9d0102v2dd.html
[17] 台灣地質資訊服務網 http://twgeoref.moeacgs.gov.tw/GipOpenWeb/wSite/ct?xItem=140862&ctNode=1233&mp=105
[18] Hyperspectral Analysis Example https://engineering.purdue.edu/~landgreb/Hyperspectral.Ex.html
[19] k-nearest neighbors algorithm
http://en.wikipedia.org/wiki/K-nearest_neighbor_algorithm
[20] E. J. Gumbel, “Bivariate Exponential Distributions” Journal of the American Statistical Association Vol. 55, No. 292 (Dec., 1960), pp. 698-707
論文全文使用權限:同意授權於2023-02-23起公開