Development Of Spacecraft Target Recognition And Pose Solution System

With the development of space technology and the gradual improvement of space station construction,rendezvous and docking of spacecraft,cabin transposition,supply of space station and other on-orbit tasks have become important contents of space plan.The primary basis for completing these space missions is to obtain the relative pose relationship between the tracking spacecraft and the target spacecraft,so as to ensure the smooth execution of the mission.Spatial relative pose relationship is usually obtained based on the visual pose measurement of the cooperative target,so the stable and accurate identification of the cooperative target is the basis for completing the mission in orbit.However,in the space environment,the cooperative target captured by optical system has many difficulties,such as poor illumination,complex background,variable target position and pose,blurred image,etc.,which brings great challenges to the stable and accurate identification of cooperative target.This paper focuses on the stable and accurate identification technology of cooperative targets in space environment.It includes the design of cooperative target,the target recognition method in complex imaging scene,the fuzzy processing method for target recognition,and the development of spacecraft target recognition and pose solution system.1.Carry out the design of cooperation targets.The pose measurement based on monocular vision firstly requires the design of a cooperative target,which determines the way and difficulty of recognition.The design of cooperative target is carried out according to the spatial working conditions.Firstly,the design principle of cooperative target is analyzed.Secondly,the important parameters of cooperative target are determined,that is,the number of cooperative target feature points and whether the feature points are in the same plane are determined.Then the specific size of the target is designed.Finally,the design results of the cooperative target are given.2.Research on cooperative target recognition methods in complex imaging scenes.In the space environment,the imaging scene is complex and there are many difficulties,which brings great challenges to the identification of cooperative targets.Firstly,an edge extraction method is designed according to the characteristics of target and environment.The method can determine the segmentation threshold adaptively and reduce the edges irrelevant to the cooperative target as much as possible to reduce the interference to the recognition.Secondly,the cooperative target is identified by the method of combination of ring feature and line feature.According to the location constraints of geometric features,the target region is determined,and then the feature points are located in this region.Then,experiments were carried out to verify the feasibility and effectiveness of the algorithm,including cooperative target edge extraction and recognition experiments,cooperative target stable and accurate recognition experiments in complex imaging scenes,and feature point positioning experiments.Experimental results show that the recognition rate of the algorithm is more than 95%.Finally,the flow of pose solution is analyzed and verified by simulation experiment.3.Research on fuzzy processing method for target recognition.Aiming at the situation that the cooperative target could not be recognized due to the fuzzy image,the improved DeblurGAN method was used to make the cooperative target data set,and the fuzzy image was restored by training and adjusting the parameters of the network.The loss function directly determines the learning objective of the network.In order to enhance the restoration effect of fuzzy images,this paper is based on the DeblurGAN network structure and adds structural similarity loss function SSIM to the loss function of the network.It overcomes the defects of pattern and line distortion and missing in original network restoration results,which is helpful for cooperative target recognition.PSNR index increased by 6.3%;The SSIM index increased by 17.6%.Through deblur processing,the restored image has sharp edges and texture details,and the restoration effect is obvious.4.Develop spacecraft target recognition and pose calculation system.In order to improve the ease of use and application value of the algorithm proposed in this paper,and efficiently identify cooperative targets and solve poses,a spacecraft target recognition and pose solution system is built.By providing graphical interface and method call interface for users,users can obtain the required information more intuitively and quickly.Firstly,the hierarchical relationship of system architecture and functional modules is designed based on the hierarchical idea.Secondly,develop each functional module to complete the construction of the whole system;Finally,the system is tested to verify the functionality and effectiveness of the developed system.