A Study Of High-Resolutopn ISAR Imaging And Its Application

The use of inverse synthetic aperture radar?ISAR?in military and civilian applications is increasing,and we want to get ISAR images with high resolution and low background noises,which facilitates subsequent target recognition and classification.Usually we can transmit a wideband signal to obtain high-range resolution.We increase the azimuth resolution by increasing the coherence time?CPI?between the target and the radar.However,it is difficult for us to track the target for a long time because of the short observation time of the radar target and the strong mobility and non-cooperation of the ISAR target.In order we can get the ISAR in real time,it has important practical significance that we can achieve the goal of high resolution imaging in short CPI.In this paper,we propose two ISAR high-resolution imaging methods based on synthetic sparse model and analytic sparse model.The two algorithms can realize high resolution ISAR imaging in short-term observation time.At the same time,the proposed algorithm has strong robustness to noise,and the background has very little nosie,which has important effect on the later target recognition.Based on the theory of compression perception,we propose a compression-aware ISAR imaging algorithm based on error back propagation.We treat the progress which transform the image data into the image field as the coding progress,the matrix which convert the data into the image domain is called coding matrix.Then we convert the image domain data to the data domain through the decoding matrix.We update the coding matrix and the decoding matrix through the error back propagation algorithm so that the error between the echo data is the smallest.Then we use the final optimized decoding matrix as a dictionary,and reconstruct the final ISAR image by using the ₀l norm method in compression perception.We introduce the cosparse analysis sparse model which has been developed in recent years into the process of ISAR imaging,analysis sparse model is a dual model of synthetic sparse model.Since the partial Fourier matrix can directly map the echo data to the Doppler domain,this is consistent with the analytic sparse model.First,we remove the phase error in echo data by the analysis dictionary's learning process.Then we use the enhanced Lagrangian algorithm to denoise the noise in the echo data.At this point,we can actually multiply the analysis dictionary and the denoised signal to obtain the ISAR image.However,in order to make the background noise in the final ISAR image substantially zero,and energy is concentrated in the target strong scattering points,we use the modified OMP algorithm to reconstruct the ISAR image;Finally,we added a multi-layer idea,and gradually improve the ISAR image resolution,so that the final ISAR image can be focused well,high resolution,and the background is almost no noise.