Research On Strip Edge Wave Detection System Based On Machine Vision

Cold-rolled strip is a high-end steel,and the shape quality is one of the most important quality indexes of cold-rolled strip,which directly affects the production efficiency and forming quality of strip.Therefore,it is of great significance to study the shape detection technology to improve the quality of cold-rolled strip products.However,shape quality defects include many types,among which edge wave defects exist in all stages of cold rolled strip production,and directly affect the smooth production and product qualification rate in subsequent stages.However,at present,the quality inspection of edge wave defects is still mainly based on manual visual measurement,which has low detection accuracy,high product objection rate and easy to cause safety hazards.In order to make up for the existing technical gap,this paper combines the machine vision detection technology to carry out the research on the strip edge wave detection system.Aiming at the transmission form of plate in cold-rolled strip production line,this paper designs a machine vision detection system composed of industrial camera and light source to realize image acquisition.According to the different characteristics of images collected when the plate is in suspension transmission and contact transmission,the edge wave contour detection algorithm and edge wave curve decomposition algorithm are proposed respectively.Firstly,the edge wave contour detection algorithm uses the seed point growth method with adaptive threshold to segment the region;Secondly,the edge wave contour is extracted by the optimized Canny edge detection algorithm.Then RANSAC is used to fit and optimize the contour curve to complete the edge wave detection.The edge wave curve decomposition algorithm first uses image rotation to calibrate the tilt,and then fits the discrete points into curve equations by Fourier series.Finally,the curve equation is decomposed by the improved VMD algorithm,the baseline drift is eliminated and the fluctuation trend histogram is obtained,from which the size of the edge wave can be judged.Experiments show that this method can realize the effective quality detection of plate edge waves under the two transmission modes of suspension and contact,and achieve the expected goal.