A Hot Rolled Steel’s Length Detection System Based On Image Rim Recognition

Iron production is vital in manufacturing industry. The quality of the industrial products is due to iron and steel’s quality. Hot rolled steel comes from pressure processing, so the rim of the steel head region will present irregular, which cannot be used. So the length detection reference should just exclude the irregular region. If the reference position is unreasonable and make the steel slab contains the irregular part, the usable length becomes shorter and lead to scrap. Conversely, the sheared length too long will make a waste. So it’s essential to abstract the slab rim, in order to, resolve the position of the length reference. By finding the corresponding relationship between pixels and real length, the steel slab’s length can be gauged, its utilization rate can be also increased.In order to meet the demand, a length detection system is designed in this paper. The system consists of control system, camera module, holder module, datum module, electric module and interface module. The system works independently at the shearing production line and can achieve a reference length of the slab, then display the detected length on the monitoring system in real-time. Operators can accomplish the shearing action based on the data and images shown on the control room’s monitor.The system has solved two thorny problems. The former one is caused by the complex circumstance in the factory. The dust and the sundries will both impact the camera, and make it difficult to abstract the rim information and interference the precision of the length. The second problem is how to use an efficient algorithm to extract the rim information from the image which has been deleted the background, and then to achieve the reference length in a satisfied precision. The algorithm should have a high running speed and stabilize to resist interference.In order to solve the first problem, in this paper we developed a composite algorithm to segment the image background and make the foreground rim information easy to pick up. The segmentation algorithm created a background model, and use the image subtracted the model. Then the image without background occurs successfully. With the help of some revise formula, the algorithm can successfully handle the thorny problem such as low luminance, reflection, shadows, and some areas with the similar colors. The image now is easy to abstract the rim information.For the sake of the solution of the second problem, two dimension Hilbert transform is put forward to abstract the rim information in this paper. Its principle is that, the gradiant variations of the signal can be amplified, so to enhance the rim information. The image’s rim information can be enhanced by extending the algorithm from one dimension into two dimensions. The algorithm can be deducted to the cotangent form by time-frequency exchange to simplify the calculation complexity. The algorithm has a high running speed and reliability and achieves a robust feature to against the complex circumstance.After getting the steel slab’s rim information, the system output an identification line to guide the shearing machine. The identification line shows the best position to cut the steel slab and remove the irregular area, then guarantee the usable length of the slab. The system is easy to installation and maintenance, its function is also safe.This hot rolled steel’s length detection system has been tested in a steel mill in Hebei province. After the installation and debugging, the system can run well and meet the steel mill’s demand, the detection precision is within±15mm. The theory is confirmed effective in this detection field, but also can be used in some other regions which are based on image processing. The application prospect is bright and vast.