Instant failure detection of steel wire rolling system using computer vision techniques

In this paper a computer vision-based method for detecting failure in production line of steel wire is proposed. Dislodging steel wire from production line is major failure in steel wire rolling systems which can cause a lot of damages in addition to workers safety risks. Our method to solve this problem is using computer vision techniques to detect the hot steel wire in real-time video captured by a fixed IP camera in front of several points of wire production line. For locating the wire in video frame to find if it is dislodged, we used an adaptive thresholding technique and compared the results with human operator reaction time. Oure experiments show that the proposed method is superior than human reaction in terms of detection speed.

کلیدواژه ها:

Computer vision ، Object detection ، Steel wire ، adaptive thresholding