Rocket Nozzle Motion Measurement System Study Based On Binocular Vision

The measurement of rocket nozzle motion parameters has great significance on the research of rocket engine thrust vectors. Ordinary contact measuring method cannot obtain rocket nozzle motion parameters with satisfactory accuracy, while vision measurement technology has advantages over non-contact and precision, which has been widely applied. Therefore, this thesis studies the vision measurement method of rocket nozzle motion attitude, and designs a rocket motion attitude measuring system based on binocular stereopsis. The main research work is as follows:Firstly, camera imaging principle is studied; a distortion correcting algorithm of image coordinates based on camera radial distortion model is derived; the calibration method of binocular cameras based on OpenCV is studied; the triangulation method is improved, and the three-dimensional reconstruction of space points is completed. In addition, the measurement of nozzle pendulum center based on dual sections and the measurement of nozzle pendulum angle based on projection are proposed.Then,the cross target with coding marks is designed in the measuring system calibration and coordinates unification section, which improves the matching speed and accuracy of corresponding target points. Several constraints are proposed, thus precisely extract and locate coding points. Since image noise affects the accuracy of decoding results, and binary coding sequence is complicated and has multiple cut-off positions, decimal decoding plan based on median filter is designed. Within a certain range of projection angles, it is capable of obtaining unique identifying information for each coding mark on the cross target. Coordinate systems of each device are unified in order to solve motion parameters. The vision measurement plan of rocket nozzle motion attitude is laid out based on semi-physical rocket nozzle model.Finally, this thesis completes the software design on VS2010 development platform. A software interface is established based on MFC. The experiment indicates that the system can extract coding points automatically, complete the decoding process, and obtain the identifying information of the coding points and their center coordinates. The calibration accuracy of binocular stereopsis system is represented by the absolute measurement error of the standard benchmark. When the system works at an approximate distance of 3600 mm, with the size of image as 1680?1710 pixels and the baseline between two cameras as 640 mm, the mean error is 0.16 mm.