側掃聲納數位影像之解析與辨識分析研究

Abstract

僅利用側掃聲納影像強度，進行海床底質的判定有其限制。必須利用聲納在線形的優勢成像以及強度具有鑑別度的條件下，進行海床底質的推估，才具有較高的正確性。 在不考量側掃聲納儀器本身的誤差，海床影像的成像強度為相對性並非絕對。並且，同一處的海床，其回散射訊號的強度，因著聲納入射角、聲波頻率與傳遞距離而有不同的反應。 配合側掃聲納掃描之成像幾何，可顯示出海床微起伏的狀態，特別是紋理、線形、陰影帶以及具有鑑別性區塊，而這些特徵可以作為影像辨識屬性的依據，並大致分類出礁石、層狀節理、礫石、砂、泥等底質。 側掃聲納於偵測波形的最佳成像方位角，為拖魚指向與波脊平行，並且坐落在測距一半以上，可增加地形起伏的辨識率。 依據相對大尺度的砂波，在拖魚指向與砂波脊線成大角度斜交時，該波形可在拖魚正下方影像顯示，並且可以辨識該波長與坡傾角，以判定此砂波為對稱波形，或非對稱波形，來指示沉積物的淨搬運方向。在向面所發展的沙條紋或大波痕也可以同時顯示於影像上。尺度相對小的波形必須採用測距相對短、拖魚與波形脊線大致平行的條件方可增加顯像對比。 依據台灣近岸的砂波影像顯示，近岸地區水下砂波的脊線走向與水流方向斜交約30-45度；在梧棲附近的砂波發展以擺線狀與貓背狀波形為主，其中擺線狀波形波長大致在200-600公尺，而貓背狀波長在400-800公尺左右，並且貓背狀波形截面積較大與主要水流方向接近正交角度斜交。梧棲附近砂波分布密度在烏溪出海口最高，並往南往北遞減，表示在此沉積物供應與搬運營力相對大。 利用側掃聲納進行海床上管線狀態的探勘，經由成像幾何的計算，可以獲得管線空間的位置狀態，並且利用地形資料的輔助更可以增加判視的正確率。管線的影像計算可利用管線前端與後端推算模式，推估裸露管線高度；並且由前側、後側的推算，可以剝離地形效應，對比實際的地形資料，得到驗證。 旋轉拖魚掃描陡坡面，在地貌探勘應用上，從水平面擴展到陡坡面。依據地形資料或影像的逆坡面頻道，估算投影坡度，以標定到正確的平面座標或者垂直座標，來顯示正確的陡坡面影像位置。 在拖魚旋轉角度與地形坡度相同以及適當的拖魚高度條件下，可達到良好的順坡面與逆坡面影像效果。在沒有地形資料條件下，利用多種旋轉角度的聲納掃描，可以將不同角度的掃描影像進行融合拼圖，以更細膩地刻劃地貌。

To discriminate seabed property has been limited by merely differentiating sonar strength of sidescan sonar. In order to achieve higher accuracy in probing seabed property, forming the image of lineation with higher sonar strength becomes a basic rule for judgment. Disregarding systematic errors, acoustic intensity of sidescan sonar image shall be compared relatively. With different incident angles, frequency and propagation length of sound wave approaching to the sea floor, different reflected scattering intensity can be shown upon the sonar image. Micro-topography, including texture, lineation, shadow zones and discriminative blocks shown on the seafloor can be displayed with reasonable sonar scan geometry. Therefore, whether the seabed has been covered by mud, sands, pebbles, rocks or reef can be differentiated. To detect wave forms lineated on the seafloor, tow fish of sidescan sonar shall be towed in the direction parallel to their strike, and the image of hinge line of sand waves shall be located in the half way of the slant range being set. When towed fish is towed in the direction perpendicular with the lineation of hinge line of sand waves, the wave length and dipping angles, can be determined from the sonar image at nadir. Under the circumstance, whether the shape of the sand wave is symmetric or not can be shown and the main direction of sediment transportation can be pointed out. In addition, those sand ribbons and megaripples developed on flank of stoss side of sand waves can be easily identified. For those smaller wave forms, a comparatively short range with the fish towed parallel to the direction of hinge line will enhance the contrast shown on the image. Side scan images of sand waves developed offshore of western Taiwan show their hinge line intersecting at 30-45 degrees with the direction of current flow. Trochoidal sand waves and cat back sand waves has been developed offshore of Wuchi. Trochoidal sand waves are about 200-600 meters in the wave length, and cat back sand waves are 400-800 meters. Cat back sand waves show with the wider cross section that is approximately perpendicular to the direction of the current flow. The denser of sand waves distributed outside the Wuchi River mouth is shown, and their appearing becomes less further to the south or north, which implies sediment supplement and transport force are larger at this region. The attitude of pipelines lying on the sea floor can be obtained by knowing their geometrical arrangement of sonar images. In addition, the estimation can be more accurate by using multi-beam bathymetrical data in the region. Regarding to the estimation of the elevated together with suspended heights of pipeline above the sea floor, they can be calculated with the Frontward and Backward models derived in this dissertation. Models are tested, and indicated the calculations have been effective. The deployment of towed fish is designed to rotate its body to scan the slope instead of the flat bottom under the water. The purpose is to observe bottom details of the bank area. In the processes, gradient of the slope can be obtained from the digital bathymetrical data or estimated from the sonar image, so that they can be projected and mapped in a space with the Cartesian coordinates. The best image to a slope area can be achieved by rotating tow fish with suitable fish heights at an angle identical with the gradient angle of the slope. Without the help of bathymetrical data, we can observe the slope by using different rotated angles, and do the mosaicking and fusion to those sonar images under the proper projection process. It can carve more detailed aspects of the slope image.