Development Of Online Inspection System For Plate Shape Based On Line Laser

The automatic control of plate shape has always been a problem to be solved in the production process of plate.With the development of CCD technology and image processing technology,the shape detection technology has been greatly improved,especially in the development of Google's deep learning framework Tensorflow and image processing OpenCV library,while promoting the development of machine vision,but also very Greatly promoted the development of technology in shape detection.Based on this,this paper studies a plate laser shape detection system based on line laser and uses TensorFlow to build a BP neural network model to realize the prediction of the width information and height information in different positions of the plate.The light stripe information on the slab surface is extracted and combined with the neural network prediction model to extract information on the height and width of the entire plate.The phase direction correlation algorithm is used to detect the motion direction,the length direction coordinates are extracted,and finally the three-dimensional reconstruction obtains the slab shape information.In the end,an on-line plate inspection system for medium and thick plates was established and the accuracy of the plate shape inspection was verified.The work of this paper helps to promote the establishment of a low-cost,high-precision industrial thick plate shape detection system,and is conducive to promoting the quality of our plate products.