Design And Implementation Of Trinocular Vision Coordinate Measuring System

Currently,visual detection system that integrates image processing,pattern recognition,and artificial intelligence technologies,is widely used in various industries for intelligent detection,because of its advantages of intelligence,non-contact,and high precision.In practice,traditional binocular vision is still affected by the camera's line of sight noise,camera performance,lens distortion,and some structural parameters.These interferences affect the axial measurement of binocular measurements,especially the accuracy and stability of the depth direction.Therefore,the trinocular vision coordinate measuring system is desigened to improve the axial accuracy and poor stability of traditional binocular vision measurement.First,we construct the measurement model of trinocular vision coordinate measurement system,design the overall composition scheme and implementation flow of the trinocular vision coordinate measurement according to the principle of the measurement system and the actual needs of the project,and select the hardware according to the project accuracy index and performance,etc.The software environment used in the development and the software implementation flow of the system are introduced.Secondly,we have studed the calibration of the camera system parameters involved in the trinocular vision coordinate measurement process,and analyzed The four coordinate systems of the camera imaging and the two linear and nonlinear imaging models.Then according to the principles of camera calibration and system calibration,the camera calibration and system calibration experiments were carried out,to obtain the system structural parameters used in the measurement process,which laid the foundation for the subsequent solution of the three-dimensional coordinates of the measured object.Then,according to the actual needs of the project,the shape of the measured object is designed and an appropriate image recognition processing algorithm is selected.The two-dimensional pixel coordinates of the measured object are solved by experimentally testing the effect of the algorithm.Next,we focus on a new method for the fusion of three-dimensional coordinates using trinocular vision fusion.It uses the correlation between the three cameras to fuse the obtained image information,the system structural parameters calibrated by the system,and the trinocular stereo vision model to achieve the solution to the three-dimensional coordinates of the target object.Divided into three steps: first to solve the initial value,and then convert the initial value to the same world coordinate system,and finally get the center of gravity of the three groups of initial values to obtain the three-dimensional coordinates of the center of the measured object.Finally,a tri-ocular vision coordinate measuring system experimental platform was built to test the scheme of this paper.We have performed experiments on the accuracy and stability of the trinocular vision coordinate measuring system,the comparison experiment of the accuracy and stability of the trinocular and binocular vision,and the experiment of the influence of the calibration accuracy on the trinocular vision coordinate measurement.The experimental results show that the trinocular vision coordinate measurement system proposed in this scheme has a maximum error of less than 4mm and a standard deviation of less than 1mm in a depth range of 3.5m and a horizontal measurement range of 1.5m-2.5m,which is better than the traditional binocular system.The accuracy and stability have been greatly improved.