Design And Research Of Spacecraft Hidden Target Point Measurement And Model Recognition

Laser lidar measurement is an advanced measurement technology developed gradually in the middle of last century.Laser lidar can be divided into mechanical Laser lidar and MEMS Laser lidar system according to the difference of components of Laser lidars.No matter what kind of the measurement system is,in principle,the position information of the measured target is obtained by transmitting high repetition rate laser pulse and capturing the return signal.MEMS laser lidar technology integrates delicately the components of Traditional laser lidar,which makes the machine more compact,low cost,easy mass production,lightweight,etc.It is easily integrated into the high-precision equipment and of high application value.Object recognition and det ection is one of the most popular research directions of artificial intelligence at this stage.At present,it has been widely used in automatic driving navigation,intelligent video surveillance and other fields.In these fields,it has a very broad application prospect to reduce the consumption of labor cost with the help of computer vision technology.Target detection algorithm mainly consists of algorithm and corresponding neural network matching with the algorithm,so the optimization of the algorithm and neural network has attracted the attention of scholars and technicians.In this paper,the main research issues are:the measurement of hidden target nodes in the trajectory of spacecraft and the detection of spacecraft target.The main work and achievem ents of this paper are as follows:1.In order to discover the hidden nodes in the trajectory of the spacecraft with directivity and measur e them accurately according to their characteristics,a method for measuring hidden targets of Spacecraft based on MEMS laser lidar measurement is proposed.The discrete radar filter is used to collect the upper position data in the Laser measurement process,and then the radar measurement performance index is obtained to promote the moderate decline of the directivity measurement coefficient.On this basis,the maximum likelihood of the eigenvalue is estimated with the aid of the motion characteristic conditions of the spacecraft target,and then the feature model of the spacecraft target is established by combining the relevant measurement indexes,so as to suppress the downward trend of the target point fetting frequency.Based on the stereo vision processing results of hidden targets,the most suitable measurement points for MEMS lidar are selected,and he proportion of hidden targets of spacecraft is calculated based on this condition.The measurement method of hidden target points of spacecraft under MEMS lidar measurement is established.With the application of MEMS lidar measurement technology,the micro motion frequency of target points incr eases,the directivity measurement coefficient decreases,and the hidden target node is effectively measured.2.In order to further improve the ability of spacecraft target recognition in space,a YOLOv3 algorithm based on the Darknet framework is proposed for spacecraft model target detection.Under the YOLOv3 algorithm,the YOLOv3 network model can be improved according to the system training results and target characteristics.and a more suitable YOLOv3-satellite network model is proposed.Firstly,we improve the original Darknet53 network structure to improve the accuracy of spacecraft data set detection;secondly,we use the spacecraft specific feature information analysis to select a more suitable anchor frame;finally,we use DIoU frame regression function to make up for the problem that the cost function is not sensitive to the image center.In order to verify the effectiveness and completeness of the detection method,the comparison and analysis of YOLOv3 satellite are carried out on the spacecraft data set.