An automatic timestamp detection technique using vision system

Abstract: As a part of the manufacturing process at the forging shop of Alleima AB in Sandviken, materials are heated in a furnace and then quenched in a quench tank. A temperature sensor in the outlet of the tank indicates the time when the temperature rises inside the quench tank. Since the cooling system is dynamic (has inertia), the temperature sensor registers a temperature rise with a delay. Considering the rapid change in the mechanical properties of hot materials, it is important to determine the exact time when the cooling begins. This project proposed a vision system-based automatic timestamp registration technique to facilitate the need. The proposed mechanism first detects the hot metal bars and then uses an event-based technique to determine the timestamp. Finally, the obtained timestamp has been compared with the temperature sensor measurement to determine the delay. This project uses the iba-system to perform the entire procedure. The video sequence is collected using ibaCapture and the image processing took place in ibaVision. The recorded timestamp is then transferred to ibaPDA for further analysis. The main goal is to combine temperature measurement and a computer vision system to automatically determine the delay registered by the temperature sensor. In addition to the proposed solution, this project analysed the image processing plug-in concept of ibaVision and its interface connection with other iba-modules such as ibaPDA and ibaCapture.