Submarine Cable Detection Based on LSR and Image Clustering

Submarine cable is an important carrier for intercontinental communication. Due to the unique nature of the usage environment, submarine cables require regular inspections to ensure normal operation. Usually, there are three methods used for underwater cable detection, including divers, fish towing, and cable robots. The drawbacks of these inspection methods include the need for a large amount of manpower, material resources, and financial resources. Because the depth of the operation is limited, with potential personnel risks. The limited detection range is not suitable for large-scale inspections and other issues. In order to solve the problems, autonomous underwater vehicles (AUVs) become popular tools to detect submarine cable. AUVs carry various detection equipment to achieve rough and precise exploration of submarine cables. Due to the weak impact of sea conditions on underwater operations, AUVs inspection of submarine cable can be suitable for the vast majority of scenarios. They can not only discover submarine cable on a large scale, but also can closely track and detect submarine cable. This paper proposes a method for AUV to autonomously identify submarine cable. Specifically, the proposed method uses Gabor filter to filter side scan sonar (SSS) images, and designs a scoring mechanism to select the corresponding largest image block as a suspected submarine cable. The submarine cable is determined using image clustering and position prior. In this paper, we conduct extemsive experiments to verify the effectiveness of the proposed algorithm.