不論是在海床地形的勘測,或是在海域範圍搜索上, 多音束測深系統已有相當廣泛地運用, 其雖能解決傳統單音束測深儀資料密集度不足之缺失, 但面對龐大資料之後續處理,卻是一個相當複雜的問題。受到商業儀器之限制,往往需配合其所附屬之應用軟體方可進行資料處理, 此造成研究資源之耗費與操作上之限制, 並導致使用者無法確切地掌握資料品質及提升施測技術。因此為期能將待處理資料自動進行品管作業,此一操作介面之開發,將是一刻不容緩之課題。

在所有資料品管篩選方法中,最為簡易且常見的方法,即以人為之判斷輸入該區域水深門檻值; 或是以近似曲線的方式擬定出門檻值。如此之篩選方法不具彈性與客觀性,且容易因人為因素而造成資料處理上的失誤。再者,人為判斷之篩選不僅需要依靠有經驗之勘測人員進行操作, 亦得耗費大量時間進行目視篩選, 因此更可突顯出自動化演算法之重要性。本研究將原始斜距與轉換後深度之資訊綜整合併後,使得原本一維資訊擴展至二維資訊。採用目前廣泛使用之階層式分群法,將此二維資料加以分群(異常值與正常值群組),並加入集中度之概念,進而達到篩選異常值之目的。

本研究所開發之資料品管操作介面以C++程式語言為架構平台,藉由此操作介面,可將多音束測深系統所獲取之資訊轉換為經、緯度及水深之三維資訊;處理結果呈現於視窗中提供即時資訊, 進而提供多項原始資料值之輸出, 以便操作者進行資料之分析與判斷。最後引入本研究所使用之自動篩選演算法,將可能為異常之水深值加以標示,以供操作者參考之用。

Multi-beamecho sounder systemhas been around nowfor some 40 years and their use in shallow waters for the last 14 years. With modern shallow water systems running at up to 9,600 soundings/second, data collection at the rate of approximately 250 million soundings/day system is possible. Processing of Multibeam Echo sounder (MBES) data is a challenging task from both hydrographic and technological perspectives. We recognize that a completely automatic system is improbable, but propose that significant benefits can still be had if we can automatically process good quality data, and highlight areas that probably need further attention.

We propose an algorithm that takes uncleaned MBES data and attempts to pick out outliers as possible as we can. The traditionalmethod that is still in use today by numerous software applications is based on a line-by-line processing approach. Automatically filtering for a depth window, by beam number, slope between points, quality flags and recently by whether the beam’s error is outside the IHO order for the survey are a number of ways in which the line-by-line approach has been speeded up. The fundamental differences between our method and the previous methods are that our algorithm does not actually delete any soundings at all and transform original one dimension information into two dimensions. Finally, we use Hierarchical Clustering to classifyMBES data into outliers and normal.

We develop the user interface formulti-beamecho sounder quality control. It provides almost the necessary tools and modules to perform a survey. Standard modules are Survey planning (track guidance lines, waypoints), channel design and 3D modeling, data acquisition, data QC and data processing/flagged. However, it will visualize the soundings to aid the decisionmaking process.

1. 緒論
1.1 前言................................1
1.2 研究動機........................2
1.3 文獻回顧........................3
2. 多音束測深系統簡介
2.1 系統原理與架構............6
2.2 資料整合........................7
2.3 測深資料特性與影響其品質之因素........9
3. 操作介面開發
3.1 3D顯示介面...................11
3.2 2D顯示介面...................16
3.3 資料篩選工具................20
4. 研究架構及相關理論
4.1 研究架構........................25
4.2 方法介紹........................33
4.2.1 密度分群法.................33
4.2.2 距離分群法.................39
4.3 分群法之運用與解析....40
5. 實測資料處理
5.1 測線區簡介....................51
5.2 資料處理........................53
5.2.1 人工辨識結果之記錄...53
5.2.2 比對結果.....................54
5.3 分析與討論....................56
6. 結語
6.1 討論................................67
6.2 結論................................70
6.3 未來展望........................75
附錄A....................................78
DGPS、船隻姿態儀、航向儀以及測深儀之資料格式介紹

[1] McCaullaghMT. and Ross C G.,”Delaunay Triangulation of a Random Data Set For IirarithmicMapping,” The Cartographic Journal, 93-99 (1980).
[2] Guenther, G.C., and Green, J.E,”Improved Depth Selection in the Swath Survey System(BS3)Combined Offline Processing (COP) Algorithm,” NOAA Technical Report, 1982.
[3] Stephen P. Matula, ”Using Exclusive Economic Zone Digital Swath Data to Select Effective Bathymetry,” National Oceanic and Atmospheric Administration, 1986.
[4] Eeg J.,”On the Identification of Spikes in Soundings,” International Hydrographic Review, 33-41 (1995).
[5] Fernando Manuel Freitas Artilheiro Diploma, ”Analysis and Procedures of Multibeam Data Cleaning for Bathymetric Charting,”Naval Military Sciences Portuguese Naval Academy, 1990.
[6] Du, Z., Well D. and Mayer, L.,”An Approach to Automatic Detection of Outliers inMultibeamEcho SounderData,”Hydrographic Journal, 19-23 (1995).
[7] 薛憲文, 「多音束測深儀之校正與施測人工魚礁之應用」, 中華民國第十九屆海洋工程研討會論文集, 1997。
[8] 劉金源, 《水中聲學-原理與應用》,國立中山大學, 1999。
[9] 楊光哲, ”淺水多音束聲納測深資料精確度及海底地形影像分析之研究”, 國立中山大學海下技術研究所碩士論文, 1999。
[10] 李晨心, ”多音束測深資料自動篩選法則之研究”, 中山大學海洋環境研究所碩士論文, 1999。
[11] Markus M. Breunig, Hans-Peter Kriegel, Raymond T. Ng ”LOF: Identifying Density-Based Local Outliers,” Proc. ACMSIGMOD2000 Int. Conf. OnManagement of Data, Dalles, TX, 2000.
[12] U.S. Army Corps ofEngineers,”Engineering and design Hydrographic surveying engineermanual,” 2001.
[13] Spiros Papadimitriou, Hiroyuki Kitawaga, Phillip B. Gibbons, and Christos Faloutsos ”LOCI: Fast Outlier Detection Using the Local Correlation Integral,” IRP-TR-02-09 July, 2002.
[14] 陳明欽, ”淺水多音束聲納測深資料的誤差分析研究”, 國立台灣大學海洋研究所碩士論文, 2002。
[15] 張逸中, 「即時測深與導航系統之研發」,海洋技術在軍事上應用(二) 海洋技術季刊,第十三卷第三期, 2003。
[16] 何智偉, ”以海底近似平面校正多音束測深系統架設參數之研究”, 國立中山大學海下技術研究所碩士論文, 2003。
[17] Brian Calder, ”Automatic Statistical Processing of Multibeam Echosounder Data,” International Hydrographic Review, Vol.4, No. 1, April, 53-68 (2003).
[18] 沈志勇, ”水下靜態目標物搜尋、偵測、辨識與定位之研究”,國立中山大學海洋環境及工程研究所碩士論文, 2005。