Abstract
The real-time measurement of head warping and lower buckling during the production process of rough-rolled slabs has long been a persistent technical problem at the production site. Currently, the detection of head warping and lower buckling in the production site relies on workers' operational experience for manual observation or measurement during machine downtime. In this paper, an online real-time detection system for the head warping and lower buckling of rough-rolled slab in hot continuous rolling based on visual detection is proposed, and a cascade filter based on morphological processing is developed, which can effectively remove the noise in the field environment and smooth the edge profile of the slab. A precise measurement and analysis method based on points and lines is proposed, which determines the precise values by subtracting the distance from the corner-point at the top of slab to the straight line at its lower edge from that between its upper and lower edges. The detection system in industrial applications has demonstrated high accuracy: detection error ≤ ±5 mm, type recognition rate ≥ 99%. Meeting on-site industrial production requirements.