Research On Position Detecting System Of Discharged Billet For Walking Beam Based On Machine Vision

At present,walking beam furnace is a heating treatment equipment which the billet must be passed through in production process.Walking beam furnace is the key process of rolled steel production line,and plays a vital role in smooth production.Therefore,in the billet automatic production process,the normal operation of walking beam furnace directly affects the nomal operation of entire steel plant.Steel's falling malfunction is common trouble,and the temperature of the heating furnace can reach more than 1000 degrees.When steel's falling malfunction occurs,the repair process is very difficult and complex,severely affect the steel production.Therefore,this thesis makes use of the existing equipment,combined with digital video recorder and computer,put forward the research on position detecting system of discharged billet for walking beam based on machine vision.The purpose of the thesis is to real-time monitor the walking beam reheating furnace and avoid accident occurred.The main contents of this thesis are as follows:according to the existing equipment of the steel plant,the hardware structure of the inspection system is designed,and the working flow of the whole detection system is expounded.Image preprocessing part:due to geometric deformation and we can not clearly see the distal billet,so perspective transformation is performed for the intercepted frame image to slove the perspective deformation.In addition,because of poor image quality,image enhancement is necessary to strengthen required characteristics,image enhancement method based on morphology is used to improve the contrast between edge and background for this problem.Edge extraction section:the morphological reconstruction is carried out first,and the reconstructed images were subjected to morphological minimization,through these two steps to reduce the interference of noise,then the watershed transform is performed to get the front edge of billet.Finally,the straight edge of the billet is extracted by Hough transform.Distance calculation section:the distance between the billet and the door can be calculated by calculationg the distance between the extracted edge and the back edge of the door straight.The software design of billet position detecting system is analyzed.Visual Studio 2012 is chosen as the development platform.Use OpenCV image processing library and VS libraries.Use c/c++ language to compile the desired image processing functions.The simulation results show that the perspective transformation is a good solution to the problem of geometric deformation.Image segmentation results are used as the evaluation criteria,it can be seen that image enhancement method based on morphology is the most suitable for this thesis.The improved watershed algorithm overcomes the noise interference and obtains the front edge of the image.By analyzing the data,the billet position is accurately and quickly obtained,real-time monitoring of billet position is realized.