Research On Recognition Of Key Component For Non-cooperative Target And Measurement Of Rotation Parameters For Spin-stabilized Target In Space

When operating a non-cooperative target on orbit in space, it is needed to recognize the key components of the target and to measure the motion parameters of the target, thereby it can provide relative navigation information for operation system. The method based on visual measurement system is widely used in such missions.In this paper, the recognition of attitude engine nozzle and the docking ring, as well as the spin-stabilized spacecraft rotation parameter measurement method are studied. The main work is as follows.According to the feature that the imaging of the attitude engine nozzle is ellipse, a least squares fitting ellipse detection method combined with elliptic geometric properties is used for recognizing spacecraft attitude engine nozzle. The method is first to extract the arc, then to do ellipse fitting in groups, and to filter results based on elliptic geometric properties. Physical simulation experiments show that this method can recognize spacecraft attitude engine nozzle accurately and continuously in the process of approaching.According to the geometric and grayscale characteristics of the docking ring typical structure, docking ring recognition method based on mathematical morphology is proposed. Firstly a multi-step mathematical morphology processing is applied to the image with different structures; followed by edge detection and line detection; and finally docking ring typical structure is located in the image by relative geometry analysis. Physical simulation experiments show that this method can recognize docking ring typical structure accurately in the process of approaching.Through simulation experiments and multi-time denoising method for the measurement of the rotation information of spin-stabilized spacecrafts, the accuracy of a visual measurement method combined with rigid body kinematics is developed. And the factors that influence the accuracy of this method are analyzed. The position and velocity vectors of feature points on the target surface are measured firstly by photogrammetric method, and then the rotation parameters of the target are calculated by rigid body motion theory. Digital simulation experiments show that the error of the visual system and the direction from which the cameras obverse the target are the main factor affecting the accuracy of the method, and the accuracy of the method can be improved by using redundant date to solve equations and filtering.