Research On Recognition Technology Of Low Small Slow Targets By Synthetic Aperture Imaging

As one of the threats to low-altitude security problems,the "low,small and slow" target is very necessary to select an appropriate detection technology.The optical synthetic aperture imaging system adopts a multi-aperture array structure,which divides the imaging tasks into various subsystems,which can not only reduce the difficulty of the manufacturing process but also achieve high-resolution imaging effects.It is a research hotspot in the field of optics.The application of optical synthetic aperture imaging technology to the detection and identification of "low,small and slow" targets has great research prospects.The research focus of this paper is the imaging of optical synthetic aperture and subsequent image processing.Selecting small unmanned aerial vehicles in the low-altitude field as the research target of detection,according to the technical indicators,the structure and imaging analysis of the synthetic aperture system are completed,the system imaging is reconstructed using the restoration algorithm,and a variety of different identification algorithms are compared to complete the "low-altitude" identification algorithm.Detection and identification of "small slow" targets.The main research contents and innovations of this paper are as follows:(1)The theoretical analysis and simulation study of the synthetic aperture system are completed from the aspects of structure and imaging.According to the imaging requirements of the detection system,the design parameters of the system were calculated,and the optimal design results of the synthetic aperture system were obtained.Through the comparative analysis of different parameters of the main characteristic factors,the calculation of the subject system parameters and detection range,and the actual assembly of the system,it is determined that the research target is a small unmanned aerial vehicle,and the synthetic aperture detection system is a ring-shaped seven-hole structure.In the field and a low altitude of 500 m,it can reach a detection range of 17 times the target size,which meets the detection design requirements.(2)An improved super-resolution restoration algorithm is proposed,which compensates the mid-frequency details of the image and improves the image contrast.Aiming at the intermediate frequency loss problem of synthetic aperture imaging,a super-resolution algorithm was used to restore the degraded image,and according to the imaging characteristics of "low,small and slow" targets,an improved super-resolution restoration algorithm based on generative adversarial network was proposed.Compared with other restoration algorithms,the improved super-resolution generative adversarial network proposed in this paper has a peak signal-to-noise ratio of up to 30 d B for image restoration,which is 10 d B higher than the original simulation image.(3)In view of the flight characteristics of "low,small and slow" and the running speed of the algorithm,the algorithm framework is studied,the problem of too large model training parameters is solved,and a lightweight target recognition network is obtained.The synthetic aperture system was used to collect the video data of the target,and the target data set was established,and the recognition effects were compared using various algorithms based on dynamic recognition and deep learning.The amount of training weight parameters of the improved algorithm in this paper is 159.74 MB,which is reduced by about 1.4 times,and the recognition accuracy is increased by 9%,reaching 96.67%,and the recognition effect of "low,small and slow" targets is better.