| The applications of Computer vision in the automotive industry is an important indicator of the level of development of the automobile manufacturing industry, in order to improve the speed and accuracy of the positioning system detection in computer vision, it has become the trend of domestic and foreign recognized which is development of real-time online testing accurate positioning of the computer vision systems. Computer vision positioning system well adapted to the needs of the production line of the automobile manufacturing industry positioning with the fast and flexible of industrial robots programmable and visual sensors to measure the characteristics of rapid non-contact.In the industrial production line of positioning, the positioning accuracy of the visual positioning system is a decisive factor of the positioning system used by automotive industry. The visual image processing technology and visual localization algorithm is the core of the machine vision positioning system, which directly determines the accuracy of the positioning system.The technology of image processing and the algorithm of templates to match the positioning of the computer vision positioning system has been Researched in this paper. Improve the Robert operator has been proposed to extract image edge features, get the edge information of image features and using template matching algorithm based on geometric primitives template matching to calculate the similarity of the location and the direction vector of the feature points as the match data. It also improved the impact due to the image brightness changes and lay the foundation to obtain a good match. In order to further improve the computational speed, hierarchical image pyramid search algorithm based on geometric primitives has been proposed to meet the requirements of industrial real-time using of computer vision positioning system. In order to improve the matching accuracy, the theory of least squares adjustment of the sub-pixel has been used to improve the level of positioning accuracy. Finally, the characteristic points of the two-dimensional images to determine the three-dimensional position of feature point formula is also been proposed, which can accurately access the body’s current position deviation. The six degrees of freedom (X, Y, and Z, the RX, RY, RZ) has been transformed by the industrial robot, and the loss of orientation correction through the PLC will be transmitted to the robot position error to achieve accurate positioning when the application of control the robot correction coating or crawl.In this paper, The hardware platform of Simulation is FAW Car Body Positioning System, and used the VC++6.0and Matlab to write the code of the image processing and fast geometric template matching localization algorithm. The simulation results of image processing and template matching demonstrate the algorithm studied in this paper is the measurement of fully keep up with the pace of production-line, and the positioning accuracy of visual positioning system is accepted by the industry required standard. |
PDF Full Text Request