Viewpoint Planning for Fuel Leak Detection in Rail-Mounted Inspection Robot

This paper proposes a fuel leak detection method designed for unmanned Antarctic energy platforms, applied to a rail-mounted inspection robot. By introducing a viewpoint planning strategy and utilizing vector geometry to calculate perspective features, the robot can automatically optimize its camera angles to capture images from the best viewpoints. Additionally, this paper employs multi-scale Retinex and high-pass filtering techniques to enhance image processing and achieve uniform lighting, significantly improving the accuracy of fuel leak detection. Experimental results demonstrate that this method exhibits excellent detection performance in simulated environments, validating its feasibility for practical applications.