Camera Image-based Outline Extraction for the Automated Grinding of Steel Outfittings in Shipyard

This paper presents a method for automating the grinding of steel outfittings in shipbuilding using camera-based image recognition technology. The study focuses on detecting the outlines of steel outfittings by processing images captured by a camera. Grayscale conversion and binarization techniques are applied to separate the object from the background, followed by morphological operations to remove noise. The extracted outline data is then converted into G-code to guide the path of the robotic grinding tool. To ensure that the outlines match the physical objects, the image and robot coordinate systems are synchronized. The system has been tested with a maximum error of only 1 mm.