A Pipeline Extraction on Forward-Looking Sonar Images Using the Self-Organizing Map

The underwater pipeline surveillance and inspection using an autonomous underwater vehicle (AUV) is the future of the oil and gas industries since the inspection using a remotely operated underwater vehicle (ROV) is very expensive. However, the successful deployment of AUVs relies on the accurate pipeline path extraction that can automatically navigate along its direction in real-time. Due to the limited visibility of the underwater environment, we proposed to use the forward-looking sonar (FLS) to overcome environment dependencies. However, the sonar images are very noisy from a nature of acoustic imaging itself. Thus, this paper mainly contributes a real-time pipeline extraction to overcome speckle noise and intensity inhomogeneity in FLS by utilizing cell-averaging constant false alarm rate (CACFAR) and the self-organizing map (SOM). In our experiment, the pipeline images data obtained from the Gulf of Thailand were used for the performance analysis where our method can accurately extract more than 89% of the dataset.