Real-time measurement of ring-rolling geometry using low-cost hardware

Ring-rolling is an industrial forming process to produce high-strength metal rings up to 6m diameter. Thick-walled cylindrical workpieces of material, typically metallic alloys, are compressed between two or more internal and external rollers and rotated until a target geometry, often outer diameter, is achieved. The process is inherently unstable and the process is often constrained and/or controlled to improve its stability. This paper presents an image processing algorithm for the measurement of ring geometry through photogrammetry in real-time. These measurements will form part of the feedback control system for a ring-rolling process. An off-the-shelf USB webcam is used to capture images of the ring during forming and the scene is controlled to maximise contrast and minimise occlusions of the ring. The image processing tasks include object identification, edge detection, outlier rejection and distortion rectification. The process has been implemented on a desktop-scale forming machine and has been shown to work at rates suitable for control the ring-rolling process using feedback.