Research On The Methods Of SAR Slow Moving Target Detection And Parameter Estimation

Synthetic aperture radar ground moving target indication(SAR-GMTI)technology plays an important role in military reconnaissance and civil traffic monitoring.It is capable of achieving the ground moving target detection,motion parameter estimation and relocation.However,the traditional airborne SAR-GMTI methods are mainly designed for the conventional airborne platforms.It is difficult for hypersonic-platform SAR to detect a slow moving target and estimate its velocity with high accuracy using the traditional methods.Therefore,it is necessary to research the methods of slow moving target detection and velocity estimation for hypersonic-platform SAR.On the other hand,due to the traditional SAR-GMTI methods cannot work for maneuvering SAR because of the complex trajectory of the platform,it is necessary to research the methods of slow moving target detection and velocity estimation for maneuvering-platform SAR.In this dissertation,for hypersonic-platform SAR,a slow moving target detection and azimuth velocity estimation method based on forward-(fore-)and backward-beam(aft-beam)SAR,a slow moving target azimuth velocity estimation method based on dual-channel fore-and aft-beam SAR,and a slow moving target detection method based on YOLO are researched and proposed.For maneuvering-platform SAR,taking the platform of uniformly accelerated linear motion as an example,a slow moving target detection and range-azimuth 2D velocity estimation method for a uniformly accelerated linear SAR is researched and proposed in this dissertation.The main contents and innovations are as follows:1.A slow moving target detection and azimuth velocity estimation method based on the fore-and aft-beam SAR is proposed.In the hypersonic-platform SAR,it is difficult for the traditional SAR-GMTI methods to detect a slow moving target and accurately estimate its azimuth velocity.Therefore,a detection and azimuth velocity estimation method based on the fore-and aft-beam SAR is proposed.First,the fore-and aft-beam SAR working mode is established and the back projection(BP)imaging model of a slow moving target is derived.Based on the approximation of the principle of stationary phase,the imaging results of the moving target can be divided into two cases: azimuth defocusing and azimuth non defocusing,and a critical azimuth velocity which induces the azimuth defocusing of the moving target can be obtained.At the same time,the SAR imaging position and position offset of the moving target in the fore-and aft-beams can be obtained by the BP imaging model.The analysis shows that there is only an azimuth position offset of the moving target in between the fore and aft-beam SAR images,and the offset is proportional to the azimuth velocity of the moving target.Then,the intensity difference of the fore-and aft-beam SAR images is used to suppress the static clutter and the constant false alarm rate detector is utilized to detect the moving target.After that,the azimuth velocity of the moving target is estimated roughly according to the azimuth position offset of the moving target,and a refocusing method based on the echo domain is proposed to further improve the accuracy of azimuth velocity estimation.The estimated azimuth velocity is embedded into the BP imaging process to realize the phase error compensation,then the remaining azimuth velocity of the moving target is calculated and the azimuth velocity is reestimated.Repeat the above refocusing steps until the remaining azimuth velocity is less than the critical azimuth velocity.Simulation results demonstrate the effectiveness of the proposed method.Compared with the traditional methods,the estimation accuracy of the proposed method can be improved by about one order of magnitude.2.A slow moving target azimuth velocity estimation method based on the dual-channel fore-and aft-beam SAR is proposed.Since the slow moving target azimuth velocity estimation method based on the fore-and aft-beam SAR has low accuracy and even failure to estimate the velocity of the weak moving target under strong clutter,a slow moving target azimuth velocity estimation method based on the dual-channel fore-and aft-beam SAR is proposed.First,a dual-channel fore-and aft-beam SAR working mode is constructed by adding one antenna channel to each beam,and the echo signal model and imaging model of a moving target by this mode are established.Then the displaced phase center antenna(DPCA)algorithm based on BP is derived,which avoids the channel registration and phase compensation and breaks the strict limitation condition of traditional DPCA algorithm.After the clutter suppression,the azimuth velocity of the moving target can be estimated roughly according to the azimuth position offset of the moving target in between the fore-and aft-beam SAR images.In order to further improve the accuracy of the azimuth velocity estimation and to solve the problem that the refocusing method in the echo domain is no longer applicable for the strong clutter,a refocusing method based on the image domain is proposed.In this method,the clutter-suppressed image is first converted to 2D wavenumber domain,and then a phase compensation factor is constructed by using the estimated azimuth velocity to compensate the phase error of the moving target.Finally,the image in the 2D wavenumber domain is inversely converted to the complex image domain to realize the refocusing,and a more accurate position offset is used to estimate the azimuth velocity of the moving target.Repeat the above refocusing steps until the difference between the two azimuth velocity estimates is less than the set threshold,thus obtaining a high-precision azimuth velocity estimation of the moving target.The simulations show that the proposed method can obtain a high-precision azimuth velocity estimation of the moving target in the case of strong clutter.3.A slow moving target detection and 2D velocity estimation method for a uniformly accelerated linear SAR is proposed.Since the traditional SAR-GMTI method is no longer applicable to the maneuvering platform,taking the uniformly accelerated linear motion platform as an example,a slow moving target detection and 2D velocity estimation of a multichannel uniform accelerated linear SAR is proposed.First,the echo signal model of a moving target in the multichannel uniformly accelerated linear SAR is established.Then,a method combining BP with velocity synthetic aperture radar(VSAR)is proposed to solve the problems of azimuth time calibration and phase compensation between channels of the maneuvering platform SAR.By accurately compensating the Doppler phase of the target and estimating the velocity frequency of the moving target in the multichannel SAR images,the method can realize the detection,real position and radial velocity estimation of a moving target for a maneuvering platform SAR.In order to further realize the high-accuracy estimation of the range-azimuth 2D velocity of the moving target,a velocity-aided BP(VA-BP)algorithm and a 2D velocity estimation method are proposed.First,the real position information of the moving target is used to obtain an imaging subspace.Then the search interval and step length of the azimuth velocity are set,and the corresponding range velocity set is obtained by using the relationship between the radial velocity and the azimuth velocity.Then,each set of components in the 2D velocity set is embedded into the VA-BP imaging process to obtain the SAR sub-image set with different focusing depths,and the clutter is separated by the VSAR method with the SAR sub-image set only containing the moving target obtained.Finally,the minimum entropy criterion is used to obtain the sub-image with the best quality,and the corresponding 2D velocity is the high-precision 2D velocity estimation.The proposed method can not only estimate the 2D velocity of the moving target,but also improve the imaging accuracy of the moving target.Simulation results demonstrate the effectiveness of the proposed method.4.A YOLO-based slow moving target detection method of the fore-and aft-beam SAR is proposed.As it is difficult for the hypersonic-platform SAR to detect a slow moving target,a detection method based on the fore-and aft-beam SAR and YOLO-v2 is proposed.First,the training set and test set for YOLO-v2 network are constructed.For the training set,the electromagnetic simulation software is used to obtain the high-precision scattering characteristics of the simulated targets under different irradiation angles.And the scattering characteristics are combined with the SAR image background information to generate the SAR echo data,and the training set of the network is obtained by SAR imaging.For the test set,the scattering characteristics of the simulated targets in the fore-and aft-beam irradiation mode are combined with their velocity information and the SAR image background information to generate the SAR echo data.The test set of the network is obtained by imaging the SAR echo data,and then,training and testing are carried out on the YOLO-v2 network.The test results show that the YOLO-v2 network is able to detect both stationary targets and moving targets in the SAR images.In order to further determine the moving target,for each pair of the fore-and aft-beam SAR image pairs,the intersection over union(IOU)of each prediction bounding box in the fore-beam image and all the prediction bounding boxes in the aft-beam image of each target is calculated to obtain an IOU vector.Finally,the moving target can be identified according to whether the IOU vector is a zero vector.The simulation results show that the proposed method can detect the slow moving target for hypersonic-platform SAR and greatly suppress false alarm and missing alarm targets.