側掃聲納系統(side-scan sonar system)所收集之海床聲納影像，其判讀與辨識工作大多以人為方式進行。目前從事影像工作的行業均朝自動化方式邁進，運用數值分析的技術來增進影像辨識效率，並降低人為辨識因人員訓練程度及長時間辨識所產生之各種誤差。由於側掃聲納影像之縫合線會影響自動化系統之辨識，可以運用特徵函數之值域來辨識水泥質魚礁，且以門檻值將縫合線及雜訊影響區域濾除。本研究以台灣常用之框型水泥礁與十字型保護礁為目標物，建置完整自動化影像辨識系統。

本研究建置之自動化影像辨識系統的辨識流程與方法如下：
(1)側掃聲納影像收集(Image Acquisition)：500kHz，斜距75m。
(2)影像特徵函數擷取(Feature Extraction)：灰度共現矩陣法。
(3)影像特徵函數分類(Classification)：非監督式貝氏分類器。
(4)影像類型判定：特徵函數熵值(Entropy)。
(5)目標物現況分析：目標物周長、面積、中心位置與數量。

本研究以高雄市茄萣魚礁區所收集之側掃聲納影像，用來當作自動化辨識系
統之驗證與最佳視窗尺寸選取。影像特徵函數包含一階參數之平均值與二階參數之熵值和同質性。本研究選用八張聲納影像，包含1至8組框型人工魚礁與十字型保護礁。以框形水泥礁為例，目標物辨識結果與原始影像比對，其真確率達79.41%。辨識系統所統計之人工魚礁數量為16~28座，而魚礁之實際數量為17座。經系統分析顯示，此類型聲納影像的最佳視窗尺寸為12×12pixels，滑動量為4pixel。

辨識系統實際應用之驗證作業，選用屏東縣枋寮魚礁區所收集之聲納影像，將原始聲納影像切割成較小區塊(2048×350像素)的方式進行。側掃聲納影像受到縫合線影響的部分，本研究採用未縫合法、二次分類法與縫合線濾除法探討此縫合線問題，最終以縫合線濾除法用來解決濾除雜訊及縫合線影響之方法較為可行。目標物現況分析結果顯示，影像中之魚礁數量為156~236座；目標物判定以特徵函數熵值大於1.45代表影像中具有可辨識之水泥質人工魚礁、小於1.35表示為砂泥質海床類型、介於1.35~1.45可能為較小型目標物或受聲學雜訊影響。

本系統在目標物辨識上先以特徵函數值域判定目標物存在狀態，若判定具有水泥質魚礁時，再以縫合線濾除法將雜訊濾除，達到無人為介入完全自動化辨識。

關鍵詞：側掃聲納、特徵函數、灰度共限矩陣、貝氏分類、熵值、同質性、平均值

Abstract
The interpretation and identification of information from the side-scan sonar imagery are mainly depended on visual observation and personal experiences. Recent studies tended to increase the identification efficiency by using numerical analysis methods. This can reduce the error that cause by the differences of observer’s experience as well as by extended time observation. The position around the center line of the slant range corrected side-scan sonar imagery might result in the degradation of the ability of numerical methods to successfully detect artificial objects. Theoretically, this problem could be solved by using a specific characteristic function to identify the existence of concrete reefs, and then filtering the noise of the central line area with a threshold value. This study was intended to develop fully automatic sonar imagery processing system for the identification of cubic concrete and cross-type protective artificial reefs in Taiwan offshore area.

The procedures of the automatic sonar imagery processing system are as follows:
(1) Image Acquisition：500kHz with slant range of 75m.
(2) Feature Extraction：grey level co-occurrence matrix (i.e., Entropy, Homogeneity and Mean)
(3) Classification：unsupervised Bayesian classifier.
(4) Object Identification：by characteristic feature (i.e., Entropy).
(5) Object’s Status Analysis：object’s circumference、area、center of mass and quantity.

This study used the sonar images collected at Chey-Ding artificial reef site in Kaohsiung City as a case study, aiming to verify the automatic sonar imagery processing system and find out the optimum window size. The image characteristic functions include one set of first order parameter (i.e., mean) and two sets of second order parameter (i.e., entropy and homogeneity). Eight(8) sonar images with 1-8 sets of cubic concrete and cross-type protective artificial reefs where used in this step. The identification efficiency of the system, in terms of the produce’s accuracy, is 79.41%. The results illustrated that there were 16~28 sets of artificial reefs being detected in this case which is comparable with the actual amount of 17 sets. Based on this investigation, the optimum window size was concluded to be 12×12 pixels with sliding size of 4 pixel.

Imagery collected at Fang-Liau artificial reef site of Pingtung County was tested. For the purpose of applicability, the original imagery (2048×2800 pixels) was divided into 8 consecutive smaller sized imagery frames(2048×350 pixels). The influence of using a two-fold classification procedure and a central filtering method to reduce the noise that caused by slant range correction were discussed. The results showed that central line filtering method is applicable. The results of object’s status analysis showed that there are 156-236 sets of reefs existed. Automatic determination of the target using the characteristic function of entropy is feasible. If the value is larger than 1.45, it represents positive identification of concrete artificial reefs. It can be classified as muddy sand seabed type if the value is smaller than 1.35. If the value is between 1.35~1.45, it illustrates the existence of a transition zone where objects of smaller in dimensions might exist.

To achieve the purpose of automatic operation, firstly, we have to identify the existence of the concrete reefs by using the specific characteristic function. Based on the result of existing concrete reefs, suture line filtering method will hence be used to filter the noise from the image information. For that all of the procedures are automatically operated without human intervention.


Key word: side-scan sonar ; characteristic function ; gray level co-occurrence matrix ; Bayesian classification ;entropy ; homogeneity ; mean

目錄
謝誌…………………………………………………………………………....….…………...…...…I
中文摘要……………………………………………………………………....….…………...…..…II
英文摘要……………...………………………………………………………….…………….……III
目錄…………………………………..………………………………………………….…………...V
圖目錄…………………………………………………………………………….…….….…....…VIII
表目錄…………………………………………….…………………………………...…….….…XIV
第一章 緒論…………………………………………………………………….………....….…….1
1-1 前言……………………………………………………………….…………………….1
1-2 研究動機與目的………………………………………………….…………………….1
1-2-1 研究動機.…………...................………………………….……........................1
1-2-2 研究目的………………………...………………………..…............................2
1-3 文獻回顧……………………………………………………………….……………….3
1-4 研究流程…………………………………………………………….………………….4
1-5 研究成果…………………………………………………………….………………….5
1-6 章節概要說明……………………………………………………….………………….5
第二章 側掃聲納系統………………………………………………….….……………………….6
2-1 側掃聲納系統組成元件……………..………..……………….……….………….…...6
2-2 側掃聲納系統概述………….………….……………………….………………….......7
2-2-1 側掃聲納系統原理.…………...…………………….....………........................7
2-2-2 聲納影像成像原理……………...…………………………..............................9
2-2-3 聲納影像的解析度…………..…….………………………............................10
2-3 側掃聲納影像資料處理…………….………….………………...………….…….….11
2-3-1 側掃聲納影像資料儲存與輸出.…………………………………..................11
2-3-2 側掃聲納影像資料拼圖與定位………………...…………............................12
2-4 全球衛星定位系統………………………………..………………………….……….12
2-5 魚礁區聲納影像…………………………………..………………………….……….13
第三章 影像自動化辨識系統………………………………….…....…………….……….……..24
3-1 側掃聲納影像收集…………………………………..………………………………..25
3-2 影像切割………………………………………………..………………......................25
3-2-1 非滑動視窗…………………………………...................................................25
3-2-2 滑動視窗...……………………………..………………..................................25
3-2-2 視窗類型選取...………………………..………………..................................26
3-3 影像分析………………………………...……………………………………….……27
3-3-1 影像像元………………....…………………...……………………….……...27
3-3-2 影像紋理..……………...………………..…………………………….……...27
3-3-3 紋理分析..……………...………………..…………………………….……...27
3-3-4 灰度共現矩陣..……….………………..…………………………..….……...28
3-4 參數的選取………………………………………………....………………….....…...29
3-4-1 一階統計量..……………...……………..…………………………….……...29
3-4-2 二階統計量..……….………………..…………………………..…….……...30
3-4-3 特徵函數選取..…………...……………..…………………………….……...30
3-4-4 特徵函數性質.…….………………..…………………………..….….……...31
3-5 影像特徵函數分類…………………………………………...……………….………32
3-5-1 貝氏分類法..…………...………………..………………………….………...32
3-5-2 貝氏分類器.…….…………………..…………………………..…..………...32
3-5-3 貝氏分類器設定……………………………………………………………...33
3-6 集群分析……………………………………………………...…………….…………33
3-6-1 集群分析的方法..…………...…………..……………………….…………...34
3-6-2 階層式集群分析.…….……………..…………………………...….………...34
第四章 系統驗證﹝一﹞: 視窗尺度……………………………...………….……….………........41
4-1 最佳視窗選取………………..………………………………………….……….........41
4-2 特徵函數設定………………………………………………………….………...........42
4-3 貝氏分類器設定………………………........................................................................42
4-4 貝氏分類器分類結果…………………........................................................................43
4-5 集群分析結果………………………............................................................................44
4-6 目標物影像分析……………………............................................................................46
4-6-1 目標物區塊…….…….……...…………..…………………………….……...46
4-6-2 目標物量化.…………….…………..…………………………..….….……...46
4-6-2 目標物定位.…….…………………..…………………………..….….……...47
第五章 系統驗證﹝二﹞: 辨識流程………………….…………………...……….……................58
5-1 研究區域概況…………………………………………..…………………..…………58
5-2 影像分割………………….………………………………………...…………………59
5-3 子影像類型………………….………….……………………………………..………59
5-4 自動化辨識系統結果………………….…………………………...…………………59
5-4-1 方法一: 未縫合法-含水體之聲納影像……………………………………..59
5-4-2 方法二: 未縫合法-排除水體之聲納影像…………………………………..60
5-4-3 方法三: 未縫合法-聲納影像修正………..………………………….……...61
5-4-4 十字型保護礁.….…………………..…………………………..….….……...62
5-4-5 框型人工魚礁…..…………………..…………………………..….….……...63
5-4-6 砂泥質海床.….…………………..…………………………..….….………...64
5-4-7 方法四: 縫合法-二次分類法………………………………………………..65
5-4-8 二次分類法-第一次分類……………………...……………………………..65
5-4-9 二次分類法-第二次分類…………………………………...………………..67
5-4-10 方法五: 縫合法-縫合線濾除法………….……………………….………....68
5-5 子影像類形判定與驗證…………………………………...……….…………………71
5-6 目標物影像分析……………………………………….………….………..…………71
第六章 系統應用範例………….……………….………………………………..…..…….….....119
6-1 影像分類流程…………………..………………………………..…………..…........119
6-2 子影像判定與驗證…………..………………………………..……...…....……..….119
第七章 討論、結論與未來研究方…………………….……………………………..…….….....125
7-1 討論…………………………………………………………………………………..125
7-2 結論…………………………………………………………………………………..125
7-3 未來研究方向………………………………………………………………………..127
參考文獻……………..…………………………………….............................................................128
附錄1…………………………….…………….....…………………...……………………...…….131
附錄2…………………………………………......…………………...……………………...…….157
附錄3…………………………………………......…………………...……………………...…….162

參考文獻
1.行政院農委會漁業署網站：http://www.fa.gov.tw/。
2.田文敏(2006)，「中華搜救總隊水下探測技術訓練課程教材」，109頁。
3.田文敏(2006)，「水下目標物搜尋作業原理與要點」，第15屆國防科技學術研討會暨95年度國防科技學術合作計畫成果發表會論文集，第147-154頁。
4.沈志勇(2005)，「水下目標物搜尋、偵測、辨識與定位之研究」，國立中山大學海洋環境及工程學系碩士論文，329頁。
5.沈育佳(1999)，「都會區空載多譜掃描影像紋理分析之研究」，國立中興大學土木工程學系研究所碩士論文，54頁。
6.林邦傑(1981)，集群分析及其應用。教育與心理研究，4，31-57。
7.林俊宏(2003)，「應用影像增加處理之邊界偵測演算法」，中華技術學院學報，第二十六期，310~317頁。
8.林榮章(1999)，「都會區多解像力遙測影像之紋理分析」，國立中興大學土木工程學系研究所碩士論文，71頁。
9.洪馨怡、蔡琬琳、蔡珊珊(2000)，「應用多變分析與類神經網路在信用評等之研究」，國立輔仁大學企業管理學系第34屆生產管理專題報告，57頁。
10.唐德誠(2002)，「灰度共現矩陣於多波段多極化SAR影像分類之研究」，國防大學中正理工學院軍事工程研究所碩士論文，114頁。
11.張坤鳳(2006)，「應用人工類神經網路於側掃聲納影像辨識之研究」，義守大學資訊管理研究所碩士論文，92頁。
12.陳浴秉(2005)，「空間域與頻率域影像紋理分析之研究」，國立中興大學土木工程學系碩士論文，86頁。
13.陳翔傑(2004)，「自動化車牌辨識系統設計」，國立中央大學電機工程研究 所碩士論文，65頁。
14.郭孟維(2008)，「應用側掃聲納影像進行海床靜態目標物自動化辨識與分析之研究」，國立中山大學海洋環境及工程學系碩士論文，136頁。
15.曹士亮(2008)，「墾丁現生珊瑚礁自動化分類與描繪之研究」，國立中山大學海洋環境及工程學系碩士論文，117頁。
16.游智翔(2001)，「多重bin層良率群集分析之研究」，國立清華大學統計學研究所碩士論文，62頁。
17.黃育仁、李建平、陳冠良、李志忠、張美萍、黃秀菊(2004)，「區域相鄰關係為基礎的影像搜尋法」，東海科學第六卷，17~41頁。
18.黃偉碩(2005)，「利用貝氏分類與因子分析法於半導體製程錯誤偵測與診斷」，中華大學科技管理學系暨科技管理研究所碩士論文，51頁。
19.賈方霈(2002)，「結合集群分析與資料視覺於低良率晶圓之成因探討」，私立元智大學工業工程與管理研究所碩士論文，44頁。
20.蔡營寬(2006)，「水下靜態目標物之偵蒐與辨識原理」，國立中山大學海洋環境及工程學系碩士論文，185頁。
21.鄭宇麟(2006)，「樹狀貝氏分類器狄氏先驗分配之合理性」，國立成功大學工業與資訊管理所碩士論文，64頁。
22.鄭勝文、邱逢琛(1997)，「水下技術概論」，國立編譯館，398頁。
23.黎偉(1998)，「紋理分析於遙測影像分類之研究」，國立中央大學土木工程研究所碩士論文，70頁。
24.蕭博元(2005)，「水下目標物搜尋作業原理與要點」，國立中山大學海洋環境及工程學系碩士論文，198頁。
25.戴昌鳳(2003)，「台灣的海洋」，遠足文化事業有限公司，215頁。
26.顏世濱(2004)，「以側掃聲納系統描繪琉球嶼海床地貌與現生珊瑚之研究」，國立中山大學海洋環境及工程學系碩士論文，160頁。
27.謝博文(2006)，「自動分類軟體在動作電未上的研究」，國立中央大學數學研究所碩士論文，54頁。
28.Alireza K. and Orlando J.H. (2002), “Color Image Retrieval Using Multispectral Random Field Texture Method”, Department of Electrical Engineering, Southern Methodist University, Dallas, Texas 75275.
29.Andrefouet, S.,Berkelmans, R.,Odriozola, L., Done, T.,Oliver, J. Muller-Karger, F. Coral Reefs (2002) “Choosing the appropriate spatial resolution for monitoring coral bleaching events using remote sensing’’.
30.Blondel, Ph., Sempere, J.C., Robigou, V. (1993), “Textural Analysis and Structure-Tracking for Geological Mapping：Applications to sonar images from Endeavour Segment, Juan de Fuca Ridge”, Proceedings Oceans'93. 18-21 Oct, pp.208~213.
31.Blondel, Ph., Parson, L.M., Robigou, V. (1998), “Textural Analysisof SideScan Sonar Imagery and Generic Seafloor Characterisation”, Proceedings OCEANS'98 Conference, Vol.1,28 Sept.-1 Oct, pp. 419~423.
32.Bridges, S., Hodges, J., Wooley, B., Karpovich, D. and Smith, G.B. (1999), “Knowledge Discovery in an Oceanographic Database”, Applied Intelligence, 11, pp. 135~148.
33.Blondel, Ph., (2000), “Automatic Mine Detection by Textural Analysis of COTS Sidescan Sonar Imagery”, International Journal of Remote Sensing., Vol 21, No.16, pp. 3115~3128.
34.Blondel, Ph., Pouliquen E.(2004) “Acoustic textures and detection of shipwreck
cargo-example of a roman ship near ELBA, ITALY’’.
35.Campbell, J. B.(2002) “Introduction to Remote Sensing, 3rd ed”, The Guilford Press, pp. 621.
36.Chellappa, R. and Kashyap, R. L. (1985), “Texture Synthesis Using 2-D Noncansal Autoregressive Models”, IEEE Tran. ASSP, 33(1), pp. 194~203.
37.Conners, R. M., Harlow, C. A. (1980), “Toward a structural Textureal Analyzer Based on Statistical Method”, Computer Graphies and Image Processing Vol.12, pp. 224~256.
38.Ehrich, R. W. and Foith, J. P. (1978), “A View of Texture Tepology and Texture Description”, Computer Graphics and Image Processing, Vol.8, pp. 174~202.
39.Haralick, R.M., Shanmugam, K. , Dinstenin I. (1973), “Texture Features for Image Classification”, IEEE Transactions on System, Vol-SMC-3, No.6 Nor 1973, pp. 610-621.
40.Johnson, H. and Helferty, M.(1990), “The Geological Interpretation of Side-scan Sonar”,Reviews of Geophysics, 28,4/November 1990, pp. 357~380.
41.Julian, K.(2000), “The TEX Method for Segmentation of Lungs from Chest CT Images by Textural Analysis of Sub-images”, University of New South Wales Department of Artificial Intelligence.
42.Kinsler, L. E., Frey, A. R., Coppens, A. B., and Sanders, J. V. (1982), “Fundamentals of Acoustics”, New York：Wiley, pp. 65.
43.Marceau, D., Howarth, P.J. and Dubois, J.M. (1989), “Automated Texture Extraction from High Spatial Resolution Satellite Imagery for Land-Cover Classification”, I. Geoscience and Remote Sensing Symposium, pp. 2765~2768.
44.Mazel, C.(1985), “Side Scan Sonar Training Manual”, Klein Associates, Inc. pp.144.
45.Reed, T.B. and Hussong, D.(1989), “Digital Image Processing Techniques for Enhancement and Classifition of SeaMARCII Side Scan Sonar Imagery”, Journal of Geophysical Research, Vol. 94, no. B6, pp. 7469~7490.
46.Weszka, J. S., Dyer C. R., Rosenfeld, A.(1976), “A ComparativeStudy of Texture Measures for Terrain Classification,” IEEETrans. on Syst., Man and Cybern., SMC, 6(4), pp. 269-285.
47.Wikantika, K., Harto, A. B., Tateishi, R., Wihartini, J. Tetuko S. S., Park, J. H.(2000), “An Investigation of Textual Characteristics Associatedwith Spectral Information for Land Use Classification,” InternationalGeoscience and Remote Sensing Symposium, pp. 2915-2917.