| With the rise of Industry 4.0 and intelligent manufacturing,under the premise of ensuring production efficiency and product quality,material flow tracking and process control refinement is the fcus of the steel enterprises.At present,the majority of plate production enterprises still use the traditional speed model or the method of cold and hot detector to track and control.However,these methods have the limitation of tracking area,resulting in the monitoring blind spots.In order to expand the coverage of steel plate tracking and improve the tracking precision,the use of machine vision to replace the original mechanical or electrical methods has become a new direction for the intelligentization of steel production.In the hot-rolled medium plate production line,the cooling bed covers the largest area and plays an important role in connecting link between the preceding and the following.Therefore,it is has great significance to realize the tracking and positioning of the plate in the cooling bed area for improve the production efficiency.Aiming at the problem of steel plate position tracking in cooling bed area,an on-line tracking and positioning system of hot rolling plate based on machine vision was designed.The method is to design the corresponding image processing algorithm for different parts of the cooling bed and establish the evaluation method for the whole process.Finally,through experimental comparison and application,it is verified that the designed processing system has a high tracking accuracy for the steel plate tracking and monitoring.The main research contents of this paper are as follows:(1)The processing effects of image denoising,enhancement and segmentation algorithms on steel plate image data are compared and studied.Gaussian filtering was selected as the filtering algorithm in this paper based on the peak signal-to-noise ratio(PSNR)value;the threshold method was determined as the segmentation algorithm in this paper based on the false edges and target contour conditions in the segmentation results;the accuracy of the processing used in the tracking and positioning algorithm in this paper was verified.It determines the entire target tracking and positioning processing flow;a full-flow evaluation system from image segmentation,target recognition to positioning was established.(2)A model of tracking and position for the steel plate based on the cooling bed roller background was established.According to the characteristics of image histogram and Gaussian filtering,a histogram processing method based on Gaussian filtering was proposed,and the influence of different filtering scales on the histogram was explored.The influence of the pempirical method on steel plate recognition under the background of roller background was explored.A threshold processing algorithm based on p-empirical nonlinear histogram was proposed;based on the segmentation results,a linear positioning algorithm based on the target density was proposed,and the influence of different density selections on the positioning results was studied.(3)A position tracking model of steel plate was established with the cooling bed asbackground.Based on the background and target characteristics of the bed,an improved Otsu cooling bed target segmentation method was proposed,and the influence of different image quality and gray distribution on the segmentation results was explored.Based on the target and background post-processing conditions,the target extraction method was established,and reasonable model parameters were determined.The steel plate target positioning algorithm based on the edge position of the steel plate is used.,it has a good positioning effect and the impact of different discarding rates on the target positioning result was studied.(4)Based on the study of plate tracking and positioning in this paper,a test system is established,and the tracking system was applicated to the actual production line.It verifies the positioning accuracy of the algorithm and the processing capacity of the system. |
PDF Full Text Request