Research Of The Key Technologies Of The Visual Easurement Network System Of Single Camera

With the rapid development of scientific technology, the requirement ofmodern production on machining precision and assembly precision measurementtechnology is more and more high, especially in the fields of aerospace militaryand civil, in most cases, requires production in the field real-time monitoring andmeasurement. In recent years, the three dimensional measurement technology hasget further development, widely applied in industrial detection, reverseengineering, robot visual, virtual reality, as well as other active real-timemeasuring process. The rising of visual measurement technology is just a newcoordinate measuring way to satisfy the urgent increasingly demand of themodern production of high-precision stereo whole field measurement technologyand equipment.Based on the analysis and research on the current3D coordinate visionmeasuring theory, the paper presents a kind of method on single camera visionmeasurement based on optical imaging principle of characteristic points. So thecamera measures the optical characteristic point on the beforehand opticalmeasuring rod in the position of multi network control points, then get the spacecoordinates of the measuring point in the position of the different networkcontrol points. And the measurement results are taken the normalization into thesame world coordinate system, then get the completed measurement result,achieve the global measurement.In order to achieve the global measurement, the paper sets up the singlecamera visual measurement network system in the first, and expounds the workprinciple of the measurement system. According to the building of the networkcoordinate system and the space geometry relationship, sets up the model ofmeasurement network system. Based on the total least squares algorithm, the paper presents the adaptive total least squares optimization algorithm to calculatethe attitude of optical characteristic point, defines the space coordinates ofoptical measurement probe according to the geometric constraint of thepre-defined optical measurement rod, then finds out the space coordinates ofmeasured point.The3D coordinates of measured point is embodied in the2D imaging planecoordinates of the optical characteristics point through the visual measuringsystem, therefore, the measurement precision of3D coordinates is affected by thesubpixel location precision of optical characteristics point’s imaging. Based onthe gray square weighted centroid localization algorithm, the paper presents thedouble cubic interpolation-gray square weighted centroid localization algorithm,increases the number of effective pixels around the optical characteristic pointimaging center, and reduces noise error through the gray square weighted,improves the imaging center positioning accuracy of optical characteristics point,realizes the accurate positioning of the optical characteristic points, And verifiesthe effectiveness of the method via comparing the experimental results.In the visual measurement, the relationship between the measured threedimensional space point and the corresponding two dimensional image point isdecided by the camera imaging geometrical model, the measurement precision ofthree dimensional space point coordinates is affected by camera parameters. Inorder to improve the overall measurement accuracy of visual measurment system,it must make the camera parameters calibrate accurately, so that the influencedegree of the parameter calibration error to the whole visual measuring systemperformance could be reduced to minimum. In the paper, the existing cameraparameter calibration methods are analyzed based on this camera parametercalibration technical requirements. It proposes a kind of virtual stereo cameraintrinsic parameters calibration methods based on particle swarm optimizationalgorithm, establishes camera imaging model through the camera internal andexternal parameters, initial value of the linear camera parameters is estimated viathe least squares algorithm, further researches and expounds the camera intrinsicparameters optimization calibration method based on particle swarm optimization algorithm. Compared with other calibration methods, it can provethat the method has higher accuracy in calibration, and satisfies the technicalrequirements of camera intrinsic parameters accurate calibration for the large sizevisual measurement system.After accurately calibrated and centroid localized, the visual measurementnetwork system need make the measuring data of much more network controlpoints taken the normalizaiton into the same coordinate system, the paper usesthe quaternion algorithm to take the normalizaiton for the measurement results inthe position of n measurement network control points, so as to achieve globaldata, and realize the whole measure. And the measurement results of singlemeasured point in the postion of multi network control points will be fused viathe weighted data fusion optimization algorithm. The simulation experimentsverify the validity of this method and prove that the establishment of the visualmeasurement network system can expand the range of measurement and improvethe accuracy of measurement.Evaluating the simulation results of calculation through the wholemeasurement of the measurement system, then defining the optimal scheme ofthe measurement system, Same time, evaluating the uncertainty evaluation andstandard deviation of the measurement results, it verifies the effectiveness andaccuracy of the single camera visual measurement network system.