Research On Computer Host Assembly Positioning Technology Based On Machine Vision

Nowadays,90% of the computer host assembly operations were completed by manual pipeline and PLC program control semi-automatic assembly line.Low degree of the automation,low precision of assembly,the unstable assembly results,and the low efficiency of assembly were general problems.In order to solve the problems existed in the assembly and positioning of computer host,such as the variety kinds,poor adaptability,low precision of repeated positioning,low production efficiency and low degree of automation,this paper studied the assembly features of CPU,memory strip and radiator in the process of computer assembly.The vision system of computer host assembly line were designed,The paper studied the global calibration algorithm of multi-camera and the geometric feature fitting algorithm of circle and line.The research of CPU,memory strip and radiator visual positioning technology was carried out according to the assembly characteristics,and the positioning technology was successfully applied to a computer host assembly production line,which was of great engineering guiding significance for the intelligent assembly process of computer host computer.The specific contents include:1.The assembly characteristics of CPU,memory and radiator in the process of computer assembly were carried out.The image characteristics of the positioning feature area were further analyzed according to the object geometry of the object and the image.the difficulties and key points to be solved in this paper were put forward according to the requirements of the actual production technology.Firstly,the geometric feature fitting algorithm was designed to meet the requirements of the positioning accuracy;Secondly,The parameters of multi-camera without overlapping visual field were calibrated.;And thirdly,A reasonable vision system component and a different light-emitting mode were selected to obtain the required high-quality image.2.The algorithm was improved and optimized on the basis of least square straight line fitting algorithm and elliptical fitting algorithm.and the corresponding improved algorithm was proposed according to the principle and geometric characteristics of the algorithm.On the basis of least square line fitting,the small residual square was given a larger weight coefficient,and the larger residual square square was given a smaller weight coefficient to improve the least square line fitting.The experimental simulation showed that the improved algorithm was superior to the traditional algorithm in accuracy.In order to solve the problem that outliers and noise points have poor anti-interference ability to the fitting results of elliptical equation in the process of fitting ellipse by traditional least square method,athreshold screening method for fitting curve was proposed.Through the screening of points,the outliers,noise points and points with low contribution rate to fitting were eliminated to improve the fitting stability and accuracy,so as to solve the problem of poor stability of the algorithm in practical engineering and high error rate.Through experimental simulation,the comparison of geometric characteristic parameters such as aspect ratio,inclination angle and so on was closer to the real edge than the traditional fitting,which improves the assembly accuracy and system stability.3.A global self-calibration method was proposed under the condition that the visual field of multi-camera did not overlap.The characteristics of the existing multi-camera calibration methods without overlapping field of view were deeply studied considering the need for multi-camera cooperation to collect images when obtaining high quality images with multiple local features.And a self-calibration method for parameter calibration based on the geometric features of the target workpiece was proposed under the condition that the multi-camera visual fields did not overlap.The calibration of three-dimensional parameters was simplified to the calibration of two-dimensional plane parameters,and the relationship between cameras was established by the geometric features of the target workpiece itself,in order to obtain the corresponding parameters,unify the coordinates of multiple cameras,and then made the output results easy to carry out.4.According to the requirements of computer mainframe assembly,the visual positioning system design of assembly line CPU installation,memory bar installation and radiator installation station was completed.Through reasonable selection of light source,lens camera,upper computer and other components of the system,the image acquisition mode and the lighting mode of each station were designed.the positioning algorithm suitable for the workpiece was selected according to the workpiece characteristics of different work stations.And the improved algorithm proposed in conclusion 2 and 3 was applied to the actual production,which improved the positioning accuracy and stability of the system,and the computer host assembly vision system was completed.Through the trial operation,each station was accurate,completed the assembly requirements stably,then,deliver to use.