| The research on space moving small target detection and tracking algorithm is the frontier research content in the field of space-based detection.The main shortcomings of the current research are the difficulty of complex undulating background suppression,the coarse background filtering strategy,and the lack of mobile background response methods.In order to solve these problems,this dissertation conducts research on related algorithms for space-based weak small moving target detection,explores effective algorithm models and key technologies for the environment with background drift<1p/f and SCR?1 to solve the problem of obtaining targets and ensuring low false alarms in strong clutter and complex background.Research results include:?1?Aiming at the problem that the traditional background suppression algorithm has poor effect on clutter suppression in images with SCR?1,a background suppression algorithm model based on non-local mean filteringwith fusion morphological operators?TPNLM?is proposed.The algorithm can enhance weak target in the environment with the lowest SCR?0.22,achieving an average signal-to-clutter ratio gain of 33.60 times,which is superior to other representative background suppression/target enhancement algorithms of the same kind.?2?Aiming at poor accuracy of single-frame target extraction in low signal-to-clutter ratio?SCR?1?environment,a"neighbor saliency map?NSM?"algorithm model is proposed.The algorithm is suitable for target detection in strong clutter environments such as clouds/land,and can obtain an average 50 times SCR gain,an average detection rate of 98%,and a false alarm rate of 1.22×10-5 or less in an environment with SCR?1?minimum SCR?0.55?,superior to other detection algorithms of the same type.?3?Aiming at the image sequence of a large amount of clutter,using the effectiveness of NSM,a joint detection model of neighborhood saliency map and temporal hypothesis test?NSM-THT?is proposed.The experimental results show that the proposed algorithm model can extract the target trajectory effectively in the sequence of SCR<1,the detection rate?0.99,and the average false alarm rate is about4.44*10-5.?4?Aiming at the problem of poor accuracy of slow target extraction,a time domain trajectory search algorithm model with measuring the difference of gray level in space domain?TDLMS-GSD-ASS?is proposed.The experimental results show that the proposed algorithm can achieve a detection probability of 99.6%when extracting slow target with speed<1p/f in the sequence of SCR<1.The average false alarm rate is8.05*10-5.?5?Aiming at the problem that drift background results in the degradation of weak target tracking performance,an iterative tracking algorithm model for time domain correlation background suppression and spatial regional gray level target recognition?TS-RGL?is proposed.The experimental results show that this algorithm model is capable of tracking targets with an average speed<5 p/f in a SCR<1 sequence with a maximum background drift rate of 0.8 p/f.In summary,the disseration gives a detailed introduction to the theory of space-based targets.In view of the shortcomings of the current research and the research difficulties of the subject,different algorithm models are proposed to solve the corresponding key problems,and detailed comparative discussion is made.In order to further verify the validity of the algorithm models,a variety of extreme scenarios have been set up for testing,and the effectiveness of the algorithm models has been fully verified.This topic provides some theoretical support for the further research of space-based small moving target detection. |
PDF Full Text Request