Study On Line Structured Light Based Visual Measurement System And Its Application

Measurement technology plays a very important role in industrial production and scientific research.Vision measurement as a new measurement method has obvious advantages over traditional measurement methods.Its main characteristics include noncontact,high accuracy,large range,simple and flexible.Line structured light vision measurement is a visual measurement method using line structured light to assist measurement.It has the advantages of high measurement accuracy,flexible measurement objects,and high measurement efficiency.This paper studies the visual measurement system based on line structured light and its application in the measurement of geometrical dimensions of mechanical parts.The main research contents include nonlinear calibration of line structured light vision sensors,efficient extraction of the center of line structured light strips,line structured light automatic segmentation and visual measurement system construction,and measurement experiments.This work can improve the measurement accuracy and efficiency of scientific research and production processes and reduce the influence of human factors in measurement activities.The main research works of this paper are as follows:(1)A non-linear calibration method for a camera and a light plane calibration method coupled with camera calibration parameters are proposed in the line structured light vision measurement system to achieve sub-pixel level calibration accuracy of the line structured light vision sensor.Among the existing camera distortion models,a camera distortion model including radial distortion and tangential distortion is selected for the industrial scene.This model contains infinite term parameters.In this paper,the first two-stage model that occupies the dominant role is selected,and a total of four coefficients are used as model parameters.For the distortion model,an objective function including 4 distortion model coefficients,8 camera internal parameters,and 3 sets of camera external parameters is established.The Levenberg-Marquardt method is used to perform nonlinear calibration of the camera internal and external parameters.Then,the light plane feature points obtained from the camera calibration results are used to fit the light plane equation.Experiments show that the overall average pixel error of the calibrated image is 0.57 pixels,which achieves high-precision calibration of line structured light sensors.(2)Aiming at the characteristic attributes of line structured light strips and the influencing factors of imaging quality in an industrial environment,an algorithm for extracting the center of light strips based on eight-neighborhood convolution filtering technology is proposed.Because the light strip image obtained by the imaging system in the industrial environment contains a lot of noise,it will seriously affect the extraction accuracy of the center of the light strips.This paper proposes a light strip center extraction algorithm based on eight neighborhood convolution filtering technology.First,the non-linear enhancement improves the contrast between different light intensities and gray levels in the region of interest,which facilitates the use of post-noise processing and gray center of gravity methods.Then use the eight-neighborhood convolution template to filter out the noise in the image and extract the center of the light strips by the gray barycenter method according to the set threshold.Finally,a polyline fitting method is used to fit the center of the light strips.The comparison with existing algorithms shows that the accuracy of the light strips extraction of this algorithm reaches the sub-pixel level.(3)Aiming at the existing light strip contour segmentation technology which is susceptible to large noise and small application range,an automatic segmentation algorithm for light strip contour based on an adaptive detection window is proposed.In order to facilitate the calculation and representation,the algorithm converts the curvature at the bending point of the outline of the light bar into a different angle for representation.For each point on the light strips,iterate through different detection windows,use the center points of the light strips in these windows to fit straight lines,and use the difference between the forward and backward difference angles of the line as an index to measure the bending degree of the light strips,and the above bending degree index is compared with a set threshold value,and finally the segment points of the outline of the light strips are automatically extracted.(4)A visual measurement system based on line structured light is constructed and applied to the measurement of geometrical dimensions of mechanical parts.In order to realize the multi-degree-of-freedom scanning of the tested parts,the scanning platform in the measurement system uses a stepping motor to drive the grinding screw and the worm gear,respectively,to achieve linear and rotary scanning movements.The motion control,image acquisition and processing of the measurement system were programmed using Lab VIEW virtual instrument software.Finally,the constructed measurement system is used to extract feature points and point cloud data on the surface of the measured object to obtain the geometric dimensions of the mechanical parts.The actual measurement of L-shaped and conical parts shows that the measurement error is less than 0.1 mm,which verifies the function and measurement accuracy of the line structured light vision measurement system.