Research On Camber Detection Algorithm Of Hot Rolled Strip Based On Machine Vision

In the process of hot rolling,the camber phenomenon of strip intermediate billet may occur due to the influence of unbalanced temperature distribution of slab,asymmetric operation of intermediate billet and deviation of rolling pressure setting.Because the hot rolling line temperature is too high,the contact measurement method is not conducive to the implementation,so this thesis aimed at the problem of strip scythe detection in hot rolling workshop,based on machine vision online detection method of scythe detection.This thesis studies and analyzes the problem of scythe visual detection,and designs a stable,accurate,fast and efficient scythe visual detection algorithm according to some problems existing in the scythe visual detection.Aiming at the problem of unstable edge detection caused by camera shaking in the traditional scythe visual detection method,this thesis proposes to install the camera remotely and extract the target area using target detection and target tracking algorithm.According to the actual distance between the camera and the target,In this area,strip intermediate billet length of 5m is cut for camber detection.Because the target is simple and easy to distinguish,the traditional target detection and tracking algorithm is sufficient to accomplish this part of the work.Since the shape of strip intermediate billet presented in the camera is like rectangle,this paper uses the target detection method based on the connected region of the largest rectangle in the target detection.However,the visual shape characteristics of strip intermediate billet are not easy to change,and there are few problems of occlusion.Therefore,when locking the region of interest of the strip intermediate billet target,it is not necessary to overcome the influence of the appearance feature,rotation,occlusion and other adverse factors,but to measure the target feature between adjacent frames by using the robust ground target tracking method.In this thesis,a scythe detection method based on edge vector is proposed to solve the misdetection problem caused by the deviation of strip intermediate billet running.In this method,the target is extracted from the pre-processed image by image binarization,and then the impurities and disconnected areas are removed by morphological operation.Then in the image,the center pixel of the image is used as the origin to establish the coordinate system,and the edge points on both sides of the intermediate blank are extracted.Then the vector between the adjacent points is obtained,and the absolute value of the unit is calculated to calculate the relationship between the adjacent unit vectors.The bending degree of the edge of the intermediate blank is obtained,and the camber is determined successively.5000 images with camber were selected as experimental samples,and experiments were carried out on Py Charm and Open CV platforms.Meanwhile,compared with the existing detection algorithms,the results showed that the success rate of camber detection was 96.32%,and the method could detect camber of strip steel more accurately.