| With the continuous improvement of labor costs and in the intelligent manufacturing as the core of the industry 4.0 and China manufacturing 2025 strategy continues to advance,the domestic manufacturing enterprises are actively developing or purchasing the corresponding automation equipment to automate the transformation of the existing production line to improve production efficiency,reduce labor cost and solve the manufacturing labor shortage etc.Machine vision is the eyes of automation equipment,through machine vision can further improve the level of automation,machine vision is more and more widely used in the automation industry,machine vision has a bright future,but also to the requirement of image processing in real-time machine vision put forward higher requirements.Based on the analysis of the current situation and development trend of machine vision and its control system,this paper constructs a real-time visual processing system based on the real-time suite KRTS of Windows operating system developed by Kithara and verify the real-time processing system.Through the analysis and comparison of common communication methods such as serial communication,USB communication and Ethernet communication,the final choice of Ethernet communication have excellent comprehensive performance,and the Ethernet communication between vision processing system and motion control system is realized by using the TCP/IP protocol API function provided by Socket.Based on the understanding of the internal operation mechanism of Socket TCP/IP protocol and the distribution of various variables,arrays and structures in the computer memory,In this paper,the data between the visual processing system and the motion control system that require data interaction is functionally classified and encapsulated into a single structure,then with the rest of need transmission variables are put together in a same structure.Socket uses this structure to transfer directly between the visual processing system and the motion control system,eliminating the serialization and deserialization of data,improving the efficiency of data communication and simplifying the data transfer protocol,making it easier to implement transmission of data.This paper improved the traditional template matching algorithm based on circular projection,by extracting the workpiece ROI image to remove the interference of the external pixels;The principle of circular template is improved to improve the utilization of the workpiece information;Through the particle swarm iteration,we can spend less time to confirm the workpiece rotation position;Using the external contour matching information of the workpiece to be measured and the template to extract and separate the ROI image of the workpiece;The ROI image compensation mechanism is provided for the workpiece which is not fully into the visual field to identify the parts and so on,to make the template matching results more reliable.Finally,this paper combines the OpenCV functions to implement and validate the template matching algorithm,and transplant it into the real-time visual processing system. |
