本文主旨在於利用影像處理技術及模糊控制自航器，以小型船舶為測試平台，探討於船舶入港過程中沿特定導航疊標線前進如遭遇固定障礙物時，經由影像處理單元計算本船與障礙物之相對方位及距離資訊，並提供給模糊自航器完成自動避碰航行。實驗之進行乃透過架設於本船船艏之旋轉Charge Coupled Device (CCD)攝影機，擷取位於障礙物(他船)之識別目標圓桶，並經由影像處理計算目標圓桶質心位置等資訊，以做為估算本船與障礙物相對方位及距離之依據。當本船完成避碰操演後，再經由原導航疊標及自航器將船舶導引回至預定航道線上。本文之實驗軟體使用MATLAB/ Simulink xPC Target環境，在影像處理方面採用HSV色彩模型，經由適當選取影像成分中H、S、V之篩選值來達到目標物辨識選定。最後於國立台灣海洋大學小艇碼頭水域進行測試，由本船之整體航行軌跡趨勢顯示本文所提之藉由影像辨識及模糊控制來達到船舶自動避碰確有其可行性。唯本船與障礙物間之距離估測精確度仍有待提昇，以確保本文所提系統於實際港區應用之安全性及可靠性。

This work is on the application of image processing technique and fuzzy logic control to the design and implementtation of a small boat-based collision avoidance system. Specifically, during the approachingmaneuver, the ownship is required to avoid an obstacle ship anchored on its way into the port, and the ownship should track its original path defined by two leading marks immediately after the avoidance maneuver. A rotatable Charged Coupled Device (CCD) mounted on the bow of the ownship is used in capturing a set of targets placed on the obstacle ship and the relative orientation and distance are used as inputs to a fuzzy controller to conduct necessary obstacle avoidance maneuvers. The Matlab/Simulink xPC target framework has been adopted and appropriate threshold values for the
HSV (Hue, Saturation, and Value) are assigned to ensure correct image acquirement. The trend of the trajectories of a series of small boat-based collision avoidance maneuvers conducted on the waterways of the NTOU small boat harbour indicates the feasibility of the proposed ship collision avoidance system basing on image processing technique and fuzzy control method. However, the accuracy of the estimated distance between the ownship and the obstacle has to be improved to render the proposed system suitable for use in a real-world harbour operation.