The Design And Implementation Of The Intelligent Correction System Of Switch Installation Based On The Image Feature Extraction

With the improvement of the automation level in industrial production, there is a great progress in the intelligence of production line, which leads to less workers in production. Due to the difference between workers and machines, a worker may get lower production efficiency and be easier to make an error than a machine during the repetitive work. Hence, a complete detection method is needed to examine the manual work, to detect and correct the mistakes in time.On the production line of an automobile factory, the panel assembly of switch buttons of the vehicle needs to be installed by workers. Due to the lack of the suitable detection method, it is hard to discover the mistake of button sequence installed by workers, which leads to the wrong panel assembly of switch buttons on the vehicle. In response to this issue, an intelligent correction system of switch installation was developed in this study. The system mainly realizes two parts of functions: The first function is to provide guidance to the workers when they install the buttons, which could ensure the correct sequence of buttons. The second function is to detect the installed panel assembly, which could prevent the error of installation.The first part of the function of the system is mainly realized by the PLC(Programmable Logic Controller). The PLC is the junction of the hardware of the system, which is used to coordinate the work between sensors and camera, and to guide the workers for installation.The second part of the function of the system is to detect the installed panel assembly by using the technologies of machine vision. The system takes pictures of the switch panel by using the industrial camera and extracts the features of images. Compare the features with the template images that are set before, then estimate the correction of the installation of the switch buttons.According to the requirements in actually production, the main jobs in this study are listed as follows:1) Design the hardware that included the worktable, PLC, camera etc. At the same time, determine the communication program of the system. 2) Get the location of the buttons and cut the images in pieces by using Match Template. Cut the image into small images that just included the signal information of a button. Then realize the pretreatment of images by using the method such as Gaussian filters, top-hat transform, binarization and region filling. 3) Calculate the features of the images based on SIFT algorithm and optimize this algorithm. Then the method of features matching was improved, which made the speed of the system have a great progress. The speed of the system has reached the requirements of the real time. 4) The software system was programmed and tested based on Emgu CV images processing library on the.NET development platform.After test by the practical experiment, the system works well and stably, and has strong robustness, which could meet the project requirements. The system has been applied into practice.