Research On Optical Identification Method Of Typical Satellite Local Components

At present,the attack,protection and on-orbit service of important space resources and facilities,mainly including satellites,have become an important development direction of space technology in the world,and satellite identification technology is one of the key links.With the maturity of approaching technology,it is possible to take close-up observations of space targets and obtain high-resolution images based on a space-based platform,which also makes a new requirement for accurate satellite identification,especially local component identification.However,there are many researches on the detection and identification of space-based targets at home and abroad,it is concentrated on long-distance target detection and on-orbit state recognition.Thus research on the identification of important local components such as satellite main body and solar panels is rarely reported.Aiming at this question,the optical characteristics together with clustering features and corresponding parameters characterization of local components was studied in this paper,and then identification method and algorithm of typical satellite local components were proposed,after that algorithm was verified through experiments.The main work and achievements are as follows:(1)The parametric representation method and clustering model of typic al satellite local components were studied.Aiming at different kinds of typical satellites,the geometric structure and optical characteristics of components were analyzed in this paper.On this basis,we extracted the clustering features to identify local components such as satellite main body,solar panels and antennas,and the corresponding parameters characterization was given at the same time.Furthermore,combined with the satellite prior knowledge,the weight of each component was set and then clustering model was established.The feature of components can be classified based on this model,which can afford the following target identification.(2)The identification methods and algorithm of satellite local components were proposed.Combined with the clustering model and typical satellite prior knowledge,and considering blurring,noise,and low contrast due to defocus,electronic noise and platform vibration,as well as geometric deformation and occlusion due to the complicated relative motion between targets and platform,we proposed a new method based on single and multi-serial images.After that,target identification algorithm was proposed.(3)Algorithm verification and applicability analysis.The identification algorithm was tested according to digital simulation experiment and ground semi-physical experiment,multi-frame sequence images with different kinds of degradation were obtained;Further,various situations such as self-occlusion,mutual-occlusion and geometric deformation were considered.Based on single and multi-serial experimental results,accuracy and applicability of the algorithm were verified.