Study On The Key Technologies Of Computer Vision And Applications In RVD And Lunar Rover Navigation

The principles and characteristics of coordinate measurement system in countries and groups with advanced aerospace technical are summarized according to the current measurement requirements of RVD(Rendezvous and Dock) and Lunar Rover in our country. A series of research is carried out on the key technologies of computer vision in RVD and Lunar Rover, such as interior and exterior parameters accurate calibration of digital camera, automatically identifying targets in RVD, center subpixel localization of circular targets calibration in RVD, densely and accurately matching algorithm for Lunar Rover Vision System, the principles and characteristics of measurement modes in RVD, path planning based on geometry in Lunar Rover. This dissertation's main work is as following:1. The paper puts forward a Two-Step method for calibrating fish-eye camera accurately. That method only needs the cloth to establish some control points without other hardware assistances. The algorithm is simple, easy comprehended and with high accuracy...etc. We make a experiment to testify the precision of the method in the paper.2. The paper puts forward a kind of method to calibrate rough elements of exterior orientation by four control points in a plane. We calculate the coordinates of three points among four points in camera station coordinate system by four control points and their image coordinate. Then we can calculate the rough transfer parameters between the camera station coordinate system and the control coordinate system by the coordinates of three points in two coordinate systems. We calculate the accurate transfer parameters by the above rough transfer parameters. We make a experiment to testify the veracity of the method in the paper.3. The calibration method of test range is deduced to calibrate the elements of exterior orientation of binocular cameras and multi-cameras in RVD and Lunar Rove navigation system.4. Two methods are deduced to calculate transfer parameters between two coordinate systems in RVD and Lunar Rove computer vision system .And we make a experiment to testify the veracity of the method in the paper.5. A method is deduced to finish image matching in Lunar Rove computer vision system that is base on epipolar image to construct pyramid image.6. Two coded targets are designed to finish image matching in RVD. And we make a experiment to identify the two coded targets by our algorithm.7. Two subpixel location algorithms of coded targets are put forward. One is subtracting image with center of mass by grayscale weight; the other is subtracting image with edge fit. A experiment is made to identify precision and stability of algorithm.8. Epipolar line matching algorithm is put forward to finish matching targets in RVD under multi-cameras model.9. Signal-camera model, double-camera model and multi-cameras model are studied. Their precision and stability are compared.10. The two-point with circul algorithm is put forward to carry out Lunar Rove's path layout. And theory of the algorithm is given detailedly.11. The theories in the paper are triumphantly used to two tasks of "nine degrees of freedom movement simulator test" and "image matching study on double cameras of detecting on moon simulator".