Character Recognition System Of Engine Cylinder Head Surface Based On Linear Structured Light

In order to solve the problem that the characters on the cylinder head surface cannot be patted and recognized,this paper designs a set of character recognition system on the cylinder head surface based on linear structural light.Based on laser triangulation,the system is divided into two parts: contour recognition system and character recognition system.The contour recognition system determines the cylinder head model by recognizing the contour.According to the model,the character recognition system takes photos above the character area to obtain the character depth map.After image enhancement and image filtering of the depth map,the deep convolutional neural network YOLOv3 is used for training and recognition of characters.The average recognition accuracy of contour is 99.9933%,and the average recognition accuracy of character is 99.6611%,which verifies the feasibility of the proposed scheme.The experimental results show that the designed system can accurately identify the cylinder head model and casting number in the production beat,and has the characteristics of strong antiinterference and high identification accuracy.The main research contents include:(1)In this paper,the requirements of the research project are analyzed.Aiming at the difficult problem of character recognition of engine cylinder head,the preliminary scheme is proposed and the actual test is carried out.The test scheme is compared and analyzed,and the preliminary scheme is finally selected.A character recognition system of engine cylinder head surface based on linear structure light is designed and built.The system uses linear structure light to scan the cylinder head surface.The visual recognition system is divided into two parts:contour recognition and character recognition.(2)In this paper,the linear structured light camera imaging principle is analyzed,designed for cylinder head contour recognition based on line structured light platform and the cylinder head surface character recognition,the contour recognition camera and character recognition camera selection,the platform to run the conveyor belt motor,inverter,encoder,photoelectric switch,selection of robot,etc.The contour recognition platform and character recognition platform are built respectively.(3)Through comparative analysis and experimental testing of different image enhancement and image filtering algorithms,it is finally determined to use gray transformation enhancement as the image enhancement algorithm and Gaussian filtering as the image filtering algorithm.The character image of cylinder head surface was taken with the linear structured light camera,and the image was processed with Halcon.The YOLOv3 algorithm is introduced and the running environment is configured.The recognition scheme based on YOLOv3 is proposed.(4)The feasibility of the proposed scheme was verified through multiple groups of experiments.First,the cylinder head profile recognition at different speeds was tested,then the effect of cylinder head surface character recognition was tested,and finally the feasibility of the scheme was verified.