Research On The Adaptive Rapid Detection Technique Of Space Targets

With the rapid development of space science and technology,the pollution of the space environment is becoming increasingly serious,which causes the demand for space target detection increasing.The mostly common space targets are various space debris,which is the remnants formed during various space activities in space.They are not controlled by human beings and distributed randomly in space areas,which interferes with the normal operation of spacecraft.Space debris is moving at high speed in space with great kinetic energy.A centimeter-sized piece of debris looks small,but it is already capable of destroying an entire satellite.Even millimeter-sized debris can cripple a satellite.Thus,to avoid collisions between spacecraft and space targets and ensure the smooth development of space activities,space targets must be detected and traced.However,in practical application,in order to reasonably and timely plan the trajectory of spacecraft,the monitoring system needs to be able to quickly analyze and process a large number of astronomical images taken under complex observation environment.Therefore,this paper chooses adaptive space target rapid detection technology for research.Contributions of this dissertation are described below:(1)Limited by the mussy observation condition in space and the image formation element structure,there are uneven distribution of gray values and smear phenomena in the background of star maps.In order to suppress the non-uniformity of background and the influence of smear effect on target detection and recognition,a bidirectional one-dimensional median filtering algorithm is presented based on the window self-adapting method.Firstly,focusing on the defects of the traditional median filtering method which is difficult to select the window size and the high cost of the median iterative method,a method of adaptive setting of the median filtering window size according to the maximum star size of the star map is proposed.Then,according to the directional characteristics of various optical overflow phenomena in astronomical images,a bidirectional one-dimensional median filtering method is proposed for background correction.The experimental results show that the proposed method can effectively suppress the complex background containing non-uniformity and smear phenomena,and play a role in signal enhancement for space targets of different shapes.(2)A large number of background stars in astronomical images will cause interference to target detection,and the energy of target signals is weak and difficult to extract.In order to solve these problems,an adaptive space target detection algorithm for single frame target detection and a target tracking algorithm based on multivariate information fusion for target trajectory correlation are proposed.Firstly,the background stars are filtered by morphological operation and frame difference operation.Then,the interference of the residual bright line on the detection performance of the target is analyzed in the process of eliminating the dispersion phenomenon.Combining with the feature of the residual bright line,the method of locating and eliminating the residual bright line in the image is proposed,which further improves the target saliency.Secondly,based on the analysis of the design defects of the threshold of the traditional inter-frame difference method,a method of adaptive setting of the segmentation threshold using Kalman filter is proposed.After the segmentation,the single frame target detection can be completed by eliminating the discrete noise.Finally,combined with the dynamic and static attributes of the target,the confidence test is carried out to detect the real trajectory of the target.Experimental results show that the target detection algorithm proposed in this paper can rapidly detect space targets with low false alarm rate and high detection rate under complex background conditions,and the target tracking algorithm constantly pruned false trajectories in the course of trajectory association,avoiding the harm caused by combinatorial dimension explosion in the course of trajectory association,and meeting the real-time requirements of the actual monitoring system.