| With the very rapid development of the steel manufacturing industry in our country,the production of medium plates has also undergone earth-shaking changes.Whether it is the production quality or production efficiency of medium and heavy plates,higher requirements are constantly being put forward.As a key link in the production process of medium plate,the cooling bed shunt detection is of great significance to improve the production efficiency of medium and heavy plate.At present,most of the cooling bed shunting monitoring tasks of medium plate manufacturers are completed by staff,and there are a lot of subjectivity problems,error problems and even safety problems in manual detection.The advantages of object detection and recognition in the scene realize the automatic detection of plate positions in the cooling bed shunt link and the interaction of upstream and downstream data,laying the foundation for the automation and efficiency improvement of the whole process of medium plate production.In this paper,a cooling bed shunt plate position detection system based on machine vision is proposed,which realizes automatic detection,tracking,interval measurement,and upstream and downstream data interaction in the cooling bed shunt scene.A large number of image processing methods such as image stitching and image filtering are used as preprocessing to remove radial distortion and tangential distortion in the camera,filter out image interference noise,and expand the scene detection range through image stitching of dual cameras;At the same time,the Deep Lab V3+ network based on deep learning semantic segmentation performs pixel-level segmentation and extraction of the outline of the medium plates on the cooling bed,and finally achieves recognition accuracy of medium plate at 97.46%.On the basis of contour recognition,by implementing the tracking measurement algorithm,the feeding information obtained by communicating with the PLC control module is matched and integrated with the objects of the vision system,so as to realize the real-time tracking of the cooling bed scene,and feed back the steel plate information to the lower-level PLC control during the unloading process.The module realizes the upstream and downstream data interaction and whole process tracking of the cooling bed system,which showed good results... |