Line Segment Length Calculation Error Research Based On Binocular Vision

Measurement based on binocular vision is an advanced non-contact measuring technology. Owing its high precision and non-invasive, it can be applied widely in many engineering fields. But in practical application, the system error accumulated will cause3D reconstruction error.In identification of crop row structure, the line segment is considered as the basic one dimensional measurement object, which is relating to the structural parameters such as height, spacing etc. In order to improve the measuring precision and accuracy of the binocular vision system, this dissertation based on the detection of length of line segment, used the statistical methods to correct system error and studied the relation between binocular vision system error and the tilt angle of the camera, distance between the camera and the target, the height of the camera, the length of line segment.First, this dissertation analyzed the sources of error from the mathematical model of binocular vision system. The multiple nonlinear regression analysis is used to find the main affecting factors of the system error, and then the errors of the system can be calculated through the regression equation.This dissertation designed18grouping experiments. And3420of experimental data were obtained by these experiments. The nonlinear regression equation of error was obtained from the experimental data by combining with statistical analysis. Using the nonlinear equation of error, the correction value of experimental data were obtained and the error was adjusted to (-5,5) mm from (-10,10) mm, which improved the precision and the accuracy of detecting the line segment using the binocular vision system.