Search

CN-122020190-A - Airship SAR fine motion compensation method based on DSP

CN122020190ACN 122020190 ACN122020190 ACN 122020190ACN-122020190-A

Abstract

The application belongs to the technical field of fine motion compensation of SAR of airship. The application provides an airship SAR fine motion compensation method based on DSP. The embodiment of the disclosure fuses a PGA and an MD algorithm, compensates low/high order errors through the PGA, accurately estimates azimuth frequency modulation rate through the MD, corrects second order phase residual errors, constructs a high-frequency vibration space-variant model, introduces a time-varying phase error estimation module, and ensures the improvement of azimuth and distance resolution. The method comprises the steps of designing a self-adaptive sub-block partitioning strategy, overlapping sub-apertures by half length, guaranteeing phase error continuity, introducing a parameter self-optimizing mechanism, and optimizing the length of the PGA/MD sub-apertures, the number of samples and the iteration times in real time by combining scene characteristics. The method comprises the steps of constructing a PGA+MD refined model, adapting to the long hole diameter characteristic, reducing the load of a DSP, designing a real-time priority resource scheduling strategy, adopting an L2 cache to store high-frequency access data, reducing delay, combining dynamic task allocation, and controlling power consumption while guaranteeing high resolution.

Inventors

  • SHEN SIHAN
  • SU TAO

Assignees

  • 西安电子科技大学

Dates

Publication Date
20260512
Application Date
20260104

Claims (9)

  1. 1. The airship SAR fine motion compensation method based on the DSP is characterized by comprising the following steps of: Step S1, performing phase error compensation based on a multi-core DSP and a PGA phase gradient self-focusing algorithm; dividing the full aperture data into a plurality of sub-aperture data with preset overlapping degree; Carrying out phase error estimation on each sub-aperture data in parallel by utilizing SAPGA sub-function algorithm so as to obtain a local phase error estimation value of each sub-aperture data; performing splicing processing on the local phase error estimated value of each sub-aperture data to generate an initial full-aperture phase error; performing phase error compensation on the full aperture data by using the initial full aperture phase error to obtain an echo signal; s2, estimating and compensating MD Doppler frequency modulation rate for the first time; setting the length of a sub-aperture and the stepping of the sub-aperture, and estimating the tone frequency of each sub-aperture in parallel by using an SAMD function algorithm, and generating a first global tone frequency by interpolation; Performing phase compensation on the echo signal by using the first global tuning frequency to obtain an MD compensated echo signal; s3, estimating and compensating MD Doppler frequency modulation rate for the second time; increasing the length of the sub-aperture and the sub-aperture stepping, and estimating the tone frequency of each sub-aperture in parallel by using an SAMD function algorithm, and generating a second global tone frequency by interpolation; and carrying out phase compensation on the echo signal subjected to the first MD compensation by using the second global tuning frequency so as to obtain target full-aperture data.
  2. 2. The airship SAR fine motion compensation method based on the DSP of claim 1, wherein the multi-core DSP is 8 cores, including core 0, core 1, core 2, core 3, core 4, core 5, core 6 and core 7, wherein core 0 is used for pre-calculating FFT/IFFT twiddle factors, distance gate sampling position vectors, azimuth frequency adjustment and azimuth time square vectors, and is stored in L2.
  3. 3. The DSP-based airship SAR fine motion compensation method according to claim 2, wherein the step of dividing the full aperture data into a plurality of sub-aperture data having a predetermined degree of overlap, comprises: Dividing each sub-aperture data into 8 blocks according to a preset rule by using a Fork-Join model, and distributing the 8 blocks to COREs 0 to 7, wherein the preset rule is core_SEC= nrn/8, each CORE calculates access offset according to CoreID, and the initial offset of each CORE processing is nan CORE_SEC Data word length CoreID, nrn are distance dimension points, and nan are azimuth dimension points.
  4. 4. A DSP-based airship SAR fine motion compensation method according to claim 3, wherein the step of estimating the phase error of each sub-aperture data in parallel by using SAPGA sub-function algorithm to obtain the local phase error estimated value of each sub-aperture data comprises: calculating sub-aperture energy by utilizing vecsum optimization functions and VecSqrt optimization functions based on the sub-aperture data, and screening strong scattering point signals with energy larger than or equal to 4 times of mean value; Performing FFT (fast Fourier transform) of 4 times of aperture length points after zero padding on each strong scattering point signal, finding a frequency spectrum peak value by using an inline C function and DSPLIB functions, calculating signal frequency, and generating a compensation factor to perform circular displacement and linear phase compensation; sequentially decrementing the iterative window based on the length of the window, and circularly processing the compensated strong scattering point signals; in each iteration, carrying out FFT and IFFT on the current strong scattering point signal, accelerating by using fftSPxSP functions and ifftSPxSP functions of the DSP, generating frequency term compensation by optimizing functions, and moving a peak value to 0 frequency and 0 phase; calculating a phase error by using an inlined C+ cyclic expansion optimized vector point multiplication function and a conjugate point multiplication function, and weighting the result according to the energy of the strong scattering point signal; After traversing all strong scattering point signals, dynamically distributing strong point weights according to sample variances by adopting a WLS-PGA strategy, processing global non-space-variant phase errors, dividing a full aperture into self-adaptive sub-blocks by adopting a LWLS-PGA strategy, obtaining phase gradients by conjugate multiplication of azimuth to adjacent sampling points, and obtaining phase error estimated values of the iteration through weighted accumulation and integration; iterating different windows in a circulating way until all windows are iterated; And performing polynomial fitting on the phase of the phase error estimated value, and removing linear and quadratic trend terms to obtain local phase error estimated values of all the sub-aperture data.
  5. 5. The DSP-based airship SAR fine motion compensation method according to claim 4, wherein the step of performing a stitching process on the local phase error estimate values of each sub-aperture data to generate an initial full-aperture phase error comprises: For the sub-aperture data of the N points, discarding each N 1 points from the beginning to the end, reserving the middle point, and taking the N 1 points as a splicing area; For the first sub-aperture, the n 2 points before unwrapping phase, the n 3 points before saving are used as a reserved area, and the n 1 points after caching are used as a splicing area; For the middle sub-aperture, unwrapping n 1 ~n 2 -1 points as an effective area, calculating the phase difference between the unwrapping n 1 ~n 2 -1 points and the splicing area of the previous sub-aperture, subtracting the correction phase of the splicing area and taking an average value after linear fitting by polyfit, updating the splicing area and storing n 1 ~n 3 -1 points; For the final sub-aperture, unwrapping N 1 -N-1 points, processing a phase difference with the middle sub-aperture, and reserving N 1 ~n 2 -1 points to obtain a spliced full-aperture phase; And sequentially performing first-order polynomial fitting to remove residual linear trend, median offset, boundary expansion and phase filtering on the spliced full-aperture phase to generate an initial full-aperture phase error.
  6. 6. The DSP-based airship SAR fine motion compensation method according to claim 5, wherein the step of compensating the full aperture data for the phase error using the initial full aperture phase error to obtain the echo signal, comprises: ensuring that all COREs are ready through SYNC0_7_ACQ, equally dividing tasks according to distance, processing core_SEC=distance dimension number/8 by each CORE, and defining a starting address Daddr of data; Invoking cosdp functions and sindp functions, and generating factors based on the initial full-aperture phase error correction array; moving full-aperture data from DDR3 to L2, completing signal and factor point multiplication through the optimized VectorDotproduct to obtain an echo signal, and writing the echo signal back to DDR3; The SYNC0_7_RLS acknowledges that all cores are complete, releasing synchronization, allowing other cores to continue processing data.
  7. 7. The DSP-based airship SAR fine motion compensation method according to claim 6, wherein the steps of setting the sub-aperture length and the sub-aperture step, estimating the tone frequency of each sub-aperture in parallel by using the SAMD function algorithm, and generating the first global tone frequency by interpolation, comprise: Initializing the sub-aperture length, sub-aperture stepping and overlapping degree, and calculating a distance segment; initializing a core 0, generating FFT/IFFT twiddle factors, and calculating MD factors; 8 COREs are parallel, and sub-aperture tasks are distributed according to core_sec= nrn/8, wherein synchronization of all CORE data is ensured through SYNCO_7_ACQ; for each sub-aperture, a SAMD function is invoked to estimate the tone frequency, and interpolation generates a first global tone frequency.
  8. 8. The DSP-based airship SAR fine motion compensation method according to claim 7, wherein the SAMD function algorithm specifically comprises: Reading parameters and setting a DMA address; Invoking an optimization function of the SAMD function, calculating average energy of the distance unit, and taking the first KNUM high-energy samples according to the descending order of energy; traversing the first KNUM high-energy samples according to sequence numbers, removing a signal quadratic term by inertial navigation speed for each sample, dividing the signal into two sub-apertures, and filling zero to 2 times of length, and accelerating FFT by DSPF_sp_ fftSPxSP; After energy normalization, vectorConjDotproduct calculates sub-aperture frequency spectrum conjugate correlation, and adds up the modulus value of the correlation result to the total correlation spectrum; After traversing, aiming at the accumulated total correlation spectrum, DSPF_sp_ maxidx finds the spectrum peak position to obtain a frequency modulation rate deviation ka_ deviate1, and refining the deviation by cubic spline interpolation; correcting the frequency modulation rate basic value based on the ka_ deviate, repeating the steps to obtain ka_ deviate2, and finally obtaining average frequency modulation rate deviation: Wherein, the Is the inclined distance between the middle distance points, Is the reference slant distance; If the average tuning rate deviation absolute value is >10, taking a boundary value, finally obtaining a first global tuning frequency ka_est=ka0+ka_ deviate.
  9. 9. The DSP-based airship SAR fine motion compensation method according to claim 8, wherein the step of phase compensating the echo signal with the first global tuning frequency to obtain the MD-compensated echo signal comprises: Core 0 collects the first global frequency modulation of all sub-apertures, and uses inertial navigation values for the frequency modulation of the first 3 and the last 3 sub-apertures; Adopting cubic spline interpolation, and generating a continuous global frequency modulation array ka_global corresponding to the all-aperture azimuth sampling points one by one according to the frequency modulation rate estimation value of the sub-aperture center; Reversely calculating more accurate airship speed according to the average value of the global frequency adjustment array ka_global, and refreshing a cache; each core generates a secondary phase compensation factor according to the ka_global array; And carrying out phase compensation on the echo signals subjected to PGA compensation in parallel to obtain the echo signals subjected to MD compensation.

Description

Airship SAR fine motion compensation method based on DSP Technical Field The embodiment of the disclosure relates to the technical field of fine motion compensation of airship SAR, in particular to a fine motion compensation method of the airship SAR based on DSP. Background The airship SAR has the advantages of low cost, flexible route, wide angle measurement range, long residence time and the like, and has obvious application potential in the fields of disaster monitoring, infrastructure detection, border inspection and the like. However, the particularity of the airship platform brings two core technical challenges, so that the high-precision real-time imaging capability of the airship platform is severely restricted, namely, the motion error accumulation caused by ① low-speed flight is seriously limited, namely, the cruising speed of the airship is 20-30km/h and is far lower than that of an airplane, the synthetic aperture time is prolonged to 20-30 times that of the airplane, the motion error is accumulated to the wavelength magnitude in long integration time, and the residual phase error after basic compensation still does not meet the high-resolution imaging requirement. ② The high-frequency vibration causes space-variant phase errors, namely radar beams are slightly changed in direction due to high-frequency angular vibration generated by wind disturbance, the space-variant phase errors are introduced to cause target defocusing and geometric distortion, the traditional compensation method fails, the image ISLR is deteriorated by 8-12dB, and the imaging quality is reduced. The traditional motion compensation mainly relies on INS/GPS combination, and calculates an oblique distance error compensation echo through position and posture information, but has inherent limitations that GPS positioning errors are difficult to meet millimeter-level compensation requirements, INS sampling frequency is difficult to capture airship high-frequency vibration, a traditional uniform linear model cannot adapt to triaxial coupling vibration, and high-frequency space-variant phase errors still remain after basic compensation, so that an image is blurred. Although the PGA algorithm can correct the residual phase error, its O (N 3) complexity makes it more than 10 seconds for the general purpose processor to process 2048×2048 point single frame data, which cannot meet the real-time requirement. Although the DSP platform has acceleration potential by virtue of the multi-core parallel architecture, the existing scheme cannot fully utilize the hardware characteristics of the DSP platform, so that the algorithm acceleration ratio is insufficient, and the real-time threshold value cannot be broken through. Current airship SAR imaging techniques have the following drawbacks: (1) The motion compensation precision is insufficient, and the requirement of high-precision SAR imaging of the airship cannot be met The sensor precision bottleneck is that INS angle error is more than or equal to 0.1 degrees, GPS positioning error is more than or equal to 1m, only coarse error of meter level is compensated, azimuth resolution is reduced, high-frequency vibration compensation is missing, INS sampling frequency cannot capture high-frequency angle shake and linear vibration of the airship, compensated image ISLR is deteriorated, defocusing is remained, non-stationary motion modeling is invalid, triaxial coupling vibration cannot be processed by a traditional uniform-speed linear model, and azimuth space-variant error is difficult to correct. (2) The fine compensation algorithm has high calculation complexity and serious real-time deficiency The algorithm complexity is increased rapidly, the maximum method complexity O (N 4) of contrast is achieved, the single frame takes more than 20 seconds, the minimum entropy method needs to execute FFT for multiple times, the memory access amount is large and real-time requirements are not met, the serial processing architecture is low-efficiency, the improved PGA method adopts serial subblock processing, the calculated amount increases with the number K of subblocks to be 2 times, the single frame processing time is long, the hardware characteristics are not utilized, the existing algorithm is not optimized for the characteristics of DSP multi-core parallelism, EDMA high-speed transmission, inline assembly instructions and the like, and the speed ratio is only 2-3 times. (3) Poor adaptability of space-variant phase error and incapability of matching vibration characteristics of airship The space variant modeling is lack of that the existing method assumes that the phase error is unchanged and cannot process target position errors caused by small angle jitter, scene adaptation dislocation is that a sub-block strategy of the spaceborne improved PGA is suitable for ionosphere slowly-varying space variant, high-frequency strong space variant of the airship is difficult to deal with, error space invariance i