Procedure for calibrating a SAR sensor

Abstract

Method for calibrating a SAR sensor with one or more receiving channels (CH1, CH2), each receiving channel (CH1, CH2) having a plurality (M, N) of receiving branches with an antenna (α 11 , ..., α 1N ; α 21 , ..., α 2M ) and with an analog-to-digital converter (ADC) for digitizing the signal received by the antenna (α 11 , ..., α 1N ; α 21 , ..., a 2M ) received SAR signal into SAR raw data (D 11 , ..., D 1N ; D 21 , ..., D 2M ) and a beamforming unit (DBF1, DBF2), wherein the beamforming unit (DBF1, DBF2) is configured to process the SAR raw data (D 11 , ..., D 1N ; D 21 , ..., D 2M ) of the plurality (M, N) of receiving branches with a weighting factor (α 1n (rg, αz), α 2m (rg, αz)), the weighted SAR raw data (D 11 , ..., D 1N ; D 21 , ..., D 2M) of the plurality (M, N) of reception branches and storing the SAR raw data obtained for a respective reception channel (CH1, CH2) in a data memory (SP1, SP2) for later processing, the method comprising the following steps: a) Adjusting the weight factors (α 1n (αz, rg), α 2m (az, rg)) of the receiving branches to obtain unweighted SAR raw data (D 11 , ..., D 1N ; D 21 , ..., D 2M ) to obtain; b) Performing a reference target analysis per receiving branch, in which the signatures of several reference targets detected by the SAR sensor in the unweighted SAR raw data (D 11 , ..., D 1N ; D 21 , ..., D 2M ) is compared with modeled signatures and a residual response of the signature is determined for each reference target in each raw data set; c) Determining a respective weight factor (α 1n (az, rg), α2m (az, rg)) for each of the plurality (M, N) of receiving branches based on the reference target analysis, wherein the determined weight factors (α 1n (az, rg), α 2m (αz, rg)) represent calibration parameters of the SAR sensor.

Description

[0001] The invention relates to a method for calibrating a SAR sensor with one or more receiving channels, wherein each receiving channel comprises a plurality of receiving branches with an antenna, an analog-to-digital converter (ADC) for digitizing the SAR signal received by the antenna into SAR raw data, and a beamforming unit. The beamforming unit is configured to weight the SAR raw data of the plurality of receiving branches with a weighting factor, combine the weighted SAR raw data of the plurality of receiving branches, and store the SAR raw data obtained for a respective receiving channel in a data memory for later processing.

[0002] During SAR satellite missions, calibrations of the deployed SAR sensors are common practice. Examples of the approaches developed for Terra-SAR-X are described in [2], for TanDEM-X in [3], for the Cosmo-SkyMed mission in [4], for the Sentinel-1 mission in [5], and for the Radarsat-2 mission in [6]. Similar calibration approaches have also been applied for the calibration of airborne SAR sensors, namely F-SAR and DBFSAR sensors, as described in [7] and [8].

[0003] A well-known approach for external calibration is described in publication [9] and EP 3 364 212 A1. This calibration approach differs from the previously mentioned techniques in that it is based on the pulse-by-pulse analysis of reference target responses in the range-compressed raw SAR data acquired by the SAR sensor. In contrast to existing methods, neither active targets nor azimuth focusing of the SAR image / SAR recording are required. The calibration approach described in [9] introduces explicit error models and derives calibration corrections through an optimization process.In particular, it explicitly considers the uncertainty in the three-dimensional positions of the antenna phase centers to ensure consistent phase and geometry between the SAR sensor's receive channels, which is crucial for multi-channel SAR applications (including Digital Beam Forming (DBF) applications). Furthermore, the optimization approach lends itself to more comprehensive and realistic modeling of error sources. For example, the proposed estimation of antenna pointing errors can derive three-dimensional roll / pitch / yaw corrections instead of the conventional elevation / azimuth estimates. It can also provide separate corrections for the transmit and receive channels if they involve different antennas.

[0004] The present invention relates to improvements of the approach described in [9], which are particularly relevant in the case of SAR sensors with integrated digital beamforming technology.

[0005] WO 2014 / 012828 A1 discloses a method for processing high-resolution spotlight SAR raw data. A range history is fitted to a hyperbolic shape suitable for efficient SAR processing. This includes a geometric correction function that numerically incorporates an actually measured orbit shape. The method considers the orbit curvature, the platform's movement during signal transmission, and the effects of the troposphere to achieve precise focusing and resolution.

[0006] CN 1 10 146 858 A discloses a method for eliminating self-interference in frequency-modulated continuous-wave radar systems. It uses baseband processing to eliminate self-interference without affecting the echo signal, improving the probability and accuracy of target detection. The method involves adaptive adjustment of reference signals using amplitude modulation and vector modulation, followed by feedback control with a fractional least-mean-square algorithm to approximate and eliminate self-interference signals. This enables automatic adaptation to operating frequency and environmental changes and provides a fast convergence rate and high self-interference elimination rate.

[0007] It is an object of the invention to provide a method, a computer program product and a device that enable improved calibration of a SAR sensor with DBF processing.

[0008] Another object of the invention is to improve the accuracy of the antenna alignment estimates obtained during calibration.

[0009] These objects are achieved by a method according to the features of claim 1, a computer program product according to the features of claim 18 and a device according to the features of claim 19. Advantageous embodiments emerge from the dependent claims.

[0010] A method for calibrating a SAR sensor with one or more receiving channels is proposed. Each receiving channel comprises a plurality of receiving branches with an antenna and an analog-to-digital converter (ADC) for digitizing the SAR signal received by the antenna into SAR raw data, as well as a beamforming unit. The beamforming unit is configured to weight the SAR raw data of the plurality of receiving branches with a weighting factor, combine the weighted SAR raw data of the plurality of receiving branches, and store the SAR raw data obtained for a respective receiving channel in a data memory for later processing.

[0011] The procedure for calibrating the SAR sensor includes the following steps: In step a), the weighting factors of the receiving branches are adjusted to obtain unweighted SAR raw data for each receiving branch.

[0012] In a step b), a reference target analysis is carried out for each receiving branch, in which the signatures of several reference targets detected by the SAR sensor in the unweighted SAR raw data are compared with modeled signatures and a residual response of the signature is determined for each reference target in each raw data set.

[0013] In a step c), a respective weighting factor is determined for each of the plurality of receiving branches based on the reference target analysis, wherein the determined weighting factors represent calibration parameters of the SAR sensor.

[0014] The proposed method enables a simple and improved calibration of a SAR sensor with any number of receive channels and integrated DBF processing. The method is based on the approach described in publication [9] and improves it with regard to applicability and accuracy for SAR sensors with digital beamforming. In order to check and, if necessary, improve the calibration of the individual receive branches, which form the input information for DBF processing in the radar sensor, access to each individual receive branch is required. With the known method, this is no longer possible because the DBF processing has already been performed in the sensor before the data is sent to a ground station. The method is based on the idea of ​​bypassing the DBF processing for the acquisition of calibration data in order to enable access to each individual receive branch for calibration.

[0015] The steps a) to c) described above are preferably performed before operating the SAR sensor. In particular, it is sufficient to perform steps a) to c) once before operating the SAR sensor.

[0016] In a practical embodiment, the weighting factors of the receive branches for acquiring calibration data are adjusted using a multiplexing operation by cyclically switching between the receive branches or receive branch groups from receive pulse to receive pulse. A receive branch group comprises a combination of signals from multiple receive branches. The multiplexing operation is preferably implemented using binary weighting factors, whereby at the time of a receive pulse, the weighting factor of one receive branch is set to 1 and the weighting factors of the remaining receive branches or receive branch groups are set to zero. This allows the multiplexing operation to be implemented without the provision of an explicit switching element.

[0017] It is also useful if the adjustment of the weighting factors includes a demultiplexing step, which results in the receiving branches or groups of receiving branches. This allows the raw SAR data determined by a receiving branch to be subjected to the reference target analysis.

[0018] According to a further advantageous embodiment, the SAR raw data from the plurality of receiving branches or receiving branch groups are subjected to respective signal processing. In an advantageous embodiment, the signal processing comprises summation and shifting of the Doppler center frequency of the SAR raw data from the plurality of receiving branches or receiving branch groups. The shifting of the center frequency followed by summation is referred to below as Doppler superposition. The adjustment of the weighting factors according to this embodiment is carried out according to the following rule: αXn(az,rg)=exp(2πj(n−1)Naz), which are: α Xn (az,rg) the weighting factor for the nth receiving branch or the nth receiving branch group, rg Integer index of the sampling point in the distance direction, starting at zero for the first sample in each received pulse, az Integer azimuth pulse index in the calibration acquisition, starting at zero for the first transmit pulse of the data acquisition, N is the number of receiving branches or receiving branch groups.

[0019] A further advantageous embodiment provides that the adjustment of the weighting factors includes a step of reversing the center frequency shift, which results in the receiving branches or receiving branch groups. This enables the reference target analysis to be performed in step b) for each receiving branch.

[0020] According to a further expedient embodiment, the beamforming unit is configured to vary the weighting factors only in the azimuth direction when performing the multiplexing operation in order to multiplex the plurality of receiving branches or receiving branch groups onto the output of the beamforming unit.

[0021] Another useful embodiment provides for the determination of correction parameters for the viewing direction of multiple antennas of at least one group of antennas of the SAR sensor. When multiple receive branches of SAR data are acquired simultaneously, the signal-to-noise ratio (SNR) in each individual receive branch is insufficient due to ambiguities and the combined effects of radar interference, instrument noise, and low antenna gain. A low SNR thus impairs the accuracy of the antenna pointing estimate obtained during calibration. The SAR signals received from multiple antennas are subsequently combined, resulting in noise reduction and, consequently, more robust and accurate antenna pointing estimates. The majority of digitized echoes transmitted and received with a specific antenna combination represent a SAR raw data channel.

[0022] Based on a reference target analysis, in which the signatures of several reference targets in the SAR raw data are compared with modeled signatures, a residual response of the signatures is determined for each reference target and each receiving branch. The residual responses are fused for each reference target across all receiving branches of the antenna group, thereby obtaining a fused residual response for each reference target. The correction parameters for the viewing direction of several antennas of at least one group of antennas of the SAR sensor are then determined from the fused residual response.

[0023] The fusion advantageously comprises summing the residual responses for each reference target in the antenna group. The fusion may additionally or alternatively comprise a weighted averaging of the residual responses for each reference target in the antenna group, with an effective antenna pattern of the antenna group being processed as the weight. The individual residual responses are advantageously fused coherently.

[0024] A further embodiment provides that the fused residual response and a fused antenna pattern of the group of antennas are processed as input variables for determining the correction parameters.

[0025] It is also advisable for the coherent superposition, which leads to the acquisition of the fused residual response, to include a precise phase calibration of all participating channels. Therefore, a calibration of the baseline and phase offset is preferably performed before calibrating the antenna alignment. Such a highly accurate calibration of the antenna baseline and phase offset is described, for example, in [9].

[0026] According to a second aspect, a computer program product is proposed which comprises instructions which, when the program is executed by a computer, cause the computer to carry out the method according to one or more embodiment variants.

[0027] According to a further aspect of the invention, a device for calibrating a SAR sensor with one or more receiving channels is proposed, wherein each receiving channel comprises a plurality of receiving branches with an antenna and with an analog-to-digital converter (ADC) for digitizing the SAR signal received by the antenna into SAR raw data, as well as a beamforming unit. The beamforming unit is configured to weight the SAR raw data of the plurality of receiving branches with a weighting factor, combine the weighted SAR raw data of the plurality of receiving branches, and store the SAR raw data obtained for a respective receiving channel in a receiving memory for later processing. The device has a processor configured to carry out the method according to one or more embodiments.

[0028] The invention is explained in more detail below using exemplary embodiments in the drawings. They show: Fig. 1 a schematic representation of a known multi-channel SAR sensor with integrated DBF technology; Fig. 2 is a schematic representation of a channel of a SAR sensor according to the invention, in which calibration data from each individual antenna is acquired by multiplexing; Fig. 3 is a schematic representation of the further processing of the data acquired with the SAR sensor according to Fig. 2 collected data; Fig. 4 is a schematic representation of one channel of a SAR sensor according to the invention, in which calibration data from all antennas are acquired simultaneously by performing signal processing using Doppler superposition; Fig. 5 a schematic representation of the further processing of the data acquired with the SAR sensor according to Fig. 4 collected data; Fig.6 is a schematic representation of one channel of a SAR sensor according to the invention, in which calibration data from all antennas are acquired simultaneously, a combined Doppler superposition of receive branches to receive branch groups and a multiplexing operation is performed for each receive branch group; Fig. 7 a schematic representation of the further processing of the data acquired with the SAR sensor according to Fig. 6 collected data; and Fig. 8 a schematic representation of the procedure for improved estimation of an antenna orientation by increasing the signal-to-noise ratio,

[0029] Fig. Figure 1 shows a schematic representation of a known multi-channel SAR sensor using DBF (Digital Beam Forming) technology. The figure illustrates the relevant hardware components of the SAR sensor and the data flow within it.

[0030] The Fig.The SAR sensor shown in Figure 1 comprises, for example, two receiving channels CH1, CH2 (generally: CHX, where X is the number of a respective receiving channel). In principle, the number of receiving channels can also be greater than 2. The SAR sensor can also comprise only a single receiving channel, without limiting its generality.

[0031] The structure of the receive channels CH1 and CH2 is identical, so for the sake of simplicity, only the first receive channel CH1 will be referred to below. The components associated with each channel are indexed using the notation "ab." The first digit "a" of the index indicates the channel (hereinafter: 1), and the second digit "b" indicates the number of a receive branch of the channel.

[0032] The receive channel CH1 comprises a plurality N of receive branches. The number of receive branches RB 11 , ..., RB 1N can be identical or different for all reception channels. In the Fig. 1, the second receiving channel CH2 has, for example, M receiving branches RB 21 , ..., RB 2M on.

[0033] Each receiving branch RB 11 , ..., RB 1N includes an antenna α 11 , ..., α 1N , an analog-to-digital converter ADC for independent digitization of the signal α 11 , ..., α 1N received SAR signal in SAR raw data D 11 , ..., D 1N . The SAR raw data D 11 , ..., D 1N are fed as input information into a beamforming unit DBF1. A memory SP1 is connected to the beamforming unit DBF1, in which data output by the beamforming unit DBF can be stored.

[0034] The beamforming unit DBF1 is designed to process the SAR raw data D 11 , ..., D 1N the majority N of receiving branches RB 11 , ..., RB 1Nwith a weighting factor 1N with a weighting factor α 1n (az,rg) and the weighted SAR raw data of the plurality N of receiving branches RB 11 , ..., RB 1N The weighted raw SAR data obtained for the receive channel CH1 are stored in the data storage SP1 for later processing.

[0035] In other words, in a known manner, for the reception channel CH1, the signals from several antennas α 11 , ..., α 1N recorded signals with time-variant, complex weighting factors α 1n (az,rg) are combined, with the raw data for each radiation shaping unit being stored in DBF1. The data stored in the data storage unit SP1 can then be transmitted via downlink for further analysis, for example, to a processing unit on Earth.

[0036] The method described below improves the external calibration technique known from [9] with regard to applicability and accuracy for multi-channel SAR sensors with integrated digital beamforming technology. Fig. 1 and the notation used in the following description largely corresponds to that used in [9] and is summarized in the following table: symbol Meaning α Xn Antenna of the receiving branch n of the receiving channel X ADC Analog-to-digital converter. Az Integer azimuth pulse index in the calibration acquisition, starting at zero for the first pulse of the data acquisition. Rg Integer range sampling index, starting at zero for the first sample in each range row. j −1 D Xn (az,rg) Complex 2D matrix containing the echoes received by antenna n. A Xn (az,rg) Complex weight of the azimuth and range variant applied to the signal received by antenna n in the associated DBF unit.

[0037] During the state-of-the-art data acquisition for SAR imaging, as described in Fig. As shown in Figure 1, the beamforming unit DBF1 of the SAR sensor combines the signals from several antennas α 11 , ..., α 1N received signals. Only the output of the beamforming unit DBF1 is stored and forwarded to the processing unit on the ground.

[0038] This creates the problem that, should the data quality prove to be poor after processing by the beamforming unit DBF1 on board the SAR sensor, a calibration correction for the receiving branches RB 11 , ..., RB 1N is required, which must be taken into account by the beamforming unit DBF1. However, the estimation of such corrections requires access to the individual reception branches into which the beamforming unit DBF1 processes, ie the raw SAR data D 11 , ..., D 1N .

[0039] In practice, calibration requires a special calibration data acquisition mode that bypasses the on-board processing by the beamforming unit DBF1 to allow calibration processing to be performed. One difficulty is that this imposes certain restrictions on the beamforming unit in the acquisition modes. For example, a maximum allowable data rate must not be exceeded.

[0040] Below, various techniques for acquiring the data required for beamforming unit calibration are described. These techniques do not increase the data rate beyond that required for the nominal imaging modes, so little or no additional hardware complexity is required to support calibration data acquisition.

[0041] The individual modes are described using a single beamforming unit DBFX of a single receive channel X. As with the usual imaging modes, the beamforming units acquire data simultaneously and in parallel. It should also be noted that the two initially described modes are special cases of a single, generalized approach, which is explained below.

[0042] The principle underlying the following examples is to use the weighting factors α Xn (az,rg) of the receiving branches RB X1 , ..., RB XN to adapt that for each receiving branch RB X1 , ..., RB XN unweighted SAR raw data D X1 , ..., D XN Subsequently, a reference target analysis is carried out for each receiving branch RB X1 , ..., RB XNcarried out in which a signature of several reference targets detected by the SAR sensor in the unweighted SAR raw data D X1 ,..., D XN compared with modeled signatures and a residual response of the signature is determined for each reference target. Then, a respective weight factor α is determined Xn (az,rg) for each of the plurality N of receiving branches RB X1 , ..., RB XN based on the reference target analysis, where the determined weight factors α Xn (az,rg) represent the desired calibration parameters of the SAR sensor.

[0043] To understand the present invention, it is sufficient to consider only the signal reception path in a SAR sensor, where processing by the beamforming unit DBFX takes place. In practice, each received signal is also connected to a transmission channel to a transmit antenna. However, the description and implementation of the method is independent of the signal transmission path, so details of the transmission are neither discussed nor included in the notation used. For example, calibration data acquisition could use the same transmission path as the usual imaging mode. While the analysis of reference targets and the derivation of calibration corrections require that the transmission path be known and characterized, these processing steps are outside the scope of the present invention.

[0044] In a first alternative, the weight factors α are adjusted Xn(az,rg) by a so-called receive branch multiplexing, which is described below using the Fig. 2 and Fig. 3 is explained in more detail.

[0045] Fig. Figure 2 shows a schematic representation of such a reception channel CHX, in which data from each individual antenna α X1 , ..., α xN captured by multiplexing. Fig. Figure 3 shows a schematic representation of the further processing of the data acquired by the SAR sensor or its receiving channel CHX after it has been stored in the data storage unit SPX. Further processing can, as described, be performed by a processing unit outside the sensor, e.g., in a data center on Earth.

[0046] In this embodiment, the processing of the beam forming unit DBFX is “replaced” by a multiplexer MUX, which is cyclically switched between the receiving branches D X1 ,..., D XNswitches from pulse to pulse and the corresponding SAR raw data D X1 ,..., D XN to the beamforming unit DBX. The multiplexing process is carried out using binary weights in the weighting factors α Xn (az,rg) of the receiving branches of the beamforming unit: αXn(az,rg)={1if az mod N=(n−1)otherwise, to multiplex the N receive branches to the output of the beamforming unit DBX.

[0047] For the actual calibration, the approach described in reference [9] is modified in such a way that before the reference target analysis ZA(D X1 , ..., D XN ) for each receiving branch D X1 ,..., D XN a step of de-multiplexing DEM is inserted, as Fig. 3. Each receiving branch resulting from the de-multiplexing corresponds to a single receiving branch RB X1 , ..., RB XNor input information for the beamforming unit DBFX. The fact that this input information has a significantly lower sampling rate is not a problem, since the target analysis described in [9] does not require SAR imaging and is therefore also capable of processing heavily subsampled data.

[0048] By multiplexing MUX, a sequence of SAR raw data is stored in the data memory SPX of the receive channel CHX in sequential order of the receive branches RB X1 , ..., RB XN This is visualized in the PSX data memory by "1", "2", ..., "N", "1", "2", ... Fig. 3 shows the downstream processing, in which the de-multiplexing DEM for each receiving branch RB X1 , ..., RB XN obtained SAR raw data are sorted and the target analysis ZA(D X1 , ..., D XN ) (target analysis).

[0049] Fig.Figure 4 shows a schematic representation of a receiving channel CHX, where calibration data from all antennas α X1 , ..., α XN are recorded simultaneously, whereby signal processing by means of Doppler superposition is carried out for all reception branches RB X1 , ..., RB XN is carried out. Fig. Figure 5 shows a corresponding schematic representation of the further processing of the data acquired by the SAR sensor or its receiving channel CHX after it has been stored in the data storage unit SPX. Further processing can, as described, be performed by a processing unit outside the sensor, e.g., in a data center on Earth.

[0050] In this embodiment, the normal processing by the beamforming unit DBFX is performed by summing the raw SAR data D received at the inputs of the beamforming unit DBFX. X1 ,..., D XN the N receiving branches RB X1 , ..., RB XNwhere a different azimuthal phase ramp SV was applied to each input information. This acquisition mode can be implemented by appropriate weighting factors in the coefficient matrices of the beamforming unit DBFX as follows: αn(az,rg)=exp(2πj(n−1)Naz), around the N input receiving branches RB X1 , ..., RB XN coherently to form an output information Σ of the beamforming unit DBFX.

[0051] For calibration, the approach described in [9] is modified in such a way that a de-ramp step ISV (see Fig. 5). In this approach, the target analysis ZA(D X1 , ..., D XN ) is applied multiple times to the same SAR raw data after they have been preprocessed with different azimuth phase ramps. The signal applied to each DBF input receive branch RB X1 , ..., RB XNis isolated by a bandpass filter, which is part of the workflow of the target analysis ZA(D X1 , ..., D XN ) (Target Analysis) as described in [9].

[0052] The alternatives for calibrating the SAR sensor described above can also be combined, as described in the Fig. 6 and Fig. 7 is shown.

[0053] During data collection, as described in Fig. 6, a Doppler superposition combination is applied independently to M groups G1,...,G M of reception branches RB X1 , ..., RB XN applied. Each of the groups G1,...,G M includes, for example, two reception branches (RB X1 , RB X2 ) ..., (RB XN-1 , RB XN). After the Doppler overlay, in which a different azimuthal phase ramp SV was applied to each input information, a multiplexer MUX switches between the M overlays to obtain the SAR raw data recorded for later calibration. For calibration, the acquired raw data are processed before the reference target analysis ZA(D X1 , ..., D XN ) is subjected to demultiplexing (DEM) and then to de-ramping ISV (see Fig. 7).

[0054] For the sake of simplicity, it was assumed in this example that the N receiving channels are divided into M equally sized groups G, which consist of M=⌈N / L⌉ comprise equally sized groups, each with L consecutive receive branches. The combined heterodyne multiplexing detection mode can be implemented by defining the weighting factors in the beamforming unit DBFX as follows: αn(az,rg)={exp(2πj(n−1)mod LL⌊azM⌋) if az mod M=⌊n−1L⌋0 otherwise

[0055] Fig. Figure 6 illustrates an example for L = 2.

[0056] The example in Fig. Figure 6 shows the application of signal processing to the raw SAR data, with the combined signals of two (generally: several) signal branches subsequently being subjected to the multiplexing step. Alternatively, the method can also first provide the multiplexing step, followed by signal processing as described above.

[0057] In the following, a further development of the described method is described by determining correction parameters for the viewing direction of several antennas of at least one group of antennas u, v, w, x, y, z of the SAR sensor described above.

[0058] The method is based on a reference target analysis ZA (Target Analysis), in which the signature of several reference targets A, B, C (generally: T) in the SAR raw data is compared with modeled signatures and for each of the reference targets A, B, C and each transmit and receive branch uv, yz (generally: st) a residual response FstT(az,rg) the signatures are determined.

[0059] The described correction of antenna pointing errors (also referred to as antenna alignment) is based on a method described in [9] and is generally applicable independently of the calibration data acquisition modes described above. Thus, the improvement is not limited to SAR sensors with integrated beamforming (DBF). It can be applied whenever it can be assumed that the antenna alignment error is identical for multiple antennas of a SAR sensor.

[0060] In the following description of determining correction parameters for the viewing direction of a SAR sensor's antennas, transmission channels must be explicitly considered. The notation used above is expanded as summarized in the following table: symbol Meaning α n The nth antenna ADC Analog-to-digital converter. Az Integer azimuth pulse index in the calibration acquisition, starting at zero for the first pulse of the data acquisition. Rg Integer range sampling index, starting at zero for the first sampling point in each range line. J −1 D st (az,rg) Complex-valued SAR raw data channel for the transmit and receive branch st, which contains the signals from the antennas α s and α t sent or received echoes. FstT(az,rg) The noise-filtered residual response (or signature) of the reference target T in the raw data D st (az,rg). ClutterstT Average residual noise energy in the residual response FstT(az,rg). FGGT(az,rg) The fused residual response of the reference target T after combining signatures from several SAR raw data channels. ClutterGGT Average residual noise energy in the fused residual response FGGT(az,rg). fr Range frequency (frequency within the bandwidth sampled by the ADC) A S (a, b, fr) Diagram of the antenna α S depending on the distance frequency and the propagation direction, parameterized by angle(α,β). AstT(az,fr) Azimuth and range-frequency dependent antenna gain to the reference target T. Derived from A S (α,β,fr) in conjunction with the channel-independent geometry of the calibration acquisition. A G (a, b, fr) Effective antenna pattern after combining signals from multiple channels.

[0061] How Fig. 1 shows, in the SAR sensor, several SAR raw data channels collect data via several receiving antennas α 11 , ..., α 1N , α 21 , ..., α 2MThe calibration approach developed in [9] is capable of estimating an independent 3D antenna pointing correction for each individual antenna. However, this is often not required in practice. Given that the individual antennas on Earth have been very carefully characterized and designed, and given the mechanical stiffness of an antenna assembly, a different antenna pointing correction for each individual antenna is usually not required.

[0062] Instead, it is proposed to apply a single, global antenna pointing error, i.e., the same antenna pointing correction to all antennas of the SAR sensor, or to determine different pointing errors for each of a comparatively small number of antenna groups. For example, the same antenna pointing correction can be applied to all antennas on the same subpanel of a planar antenna after it has been deployed in orbit.

[0063] In practice, the accuracy of antenna alignment calibration is primarily limited by the signal-to-noise ratio (SNR). Fig. As illustrated in Figure 8, the signal-to-noise ratio can be improved by using residual responses RES(FuvT),RES(FwxT),RES(FyzT),T∈A,B,C the same reference target A, B, C (second “column” in Fig. 2, where ZA(Duv),ZA(Dwx),ZA(Dyz) The first column represents the reference target analysis of the respective SAR raw data channels), which were simultaneously acquired via several antennas u, v, w, x, y, z, are additively combined. This additive combination is shown in the second and third columns of the Fig. 8. A single estimation of the gaze direction KPPD is then performed in a step of orientation estimation (last “column” in Fig. 8) was determined.

[0064] A set of transmit and receive branches C = {uv, wx, ...}, used to record SAR raw data channels for determining correction parameters (calibration), includes all antennas of at least one group G. The same antenna u, v, w, x, y, z can be used in multiple channels C, but each antenna u, v, w, x, y, z must occur at least once. Furthermore, it is useful to record the channels C simultaneously or, if multiple transmit antennas are used, as close in time as possible by interleaving different transmit pulses in a repeating pulse sequence. This simplifies processing because it ensures that the geometry of all channels C can be considered identical for antenna alignment estimation.

[0065] In addition to these general requirements, specific details may make a particular set of transmit and receive branches C preferable to others.

[0066] For example, a global alignment correction for an interferometer with two antennas, 1 and 2, such that G = {1,2}, can be derived with channels C = {11,12} or with C = {11,22}. In practice, the first set of channels, C = {11,12}, has the advantage of using only a single transmitting antenna, antenna 1, so that all channels can be received simultaneously and, consequently, a higher sampling rate can be used. This example shows that a radar system with two antennas can be calibrated in two different ways. In the first case, one always transmits with antenna 1 and receives simultaneously on antennas 1 and 2 (C = {11,12}). In the second case, one transmits and receives with antenna 1 (-> 11) and then does the same with antenna 2 (-> 22).

[0067] In another example, a polarimetric SAR instrument uses two antennas G = {1,2} for transmitting and receiving horizontally and vertically polarized signals, respectively. In this case, it may be necessary to use C = {11,22} as the set of channels to facilitate global alignment calibration, since certain commonly used reference target types are only detectable in co-polarized channels (i.e., 11 or 22) but not in cross-polarized channels (i.e., 12 or 21).

[0068] The method described here is based on and extends the methods described in Chapter 4.2 of Reference [9], formulating the method as an optimization problem that requires the following input information for all transmit and receive branches st ∈ C and reference targets T: - FstT(az,rg): The residual response of a single reference target T obtained by correcting for systematic fluctuations and noise filtering. - ClutterstT: the average residual noise energy in the filtered response RES(FstT(az,rg)) - A S (α, ß, fr) and A t (α, β, fr) as antenna patterns of the transmitting and receiving antennas.

[0069] The content of Chapter 4.2 of reference [9] is incorporated into the present application by reference.

[0070] With regard to the actual optimization problem, as formulated in equation (27) of [9], the size δRCSstT(az) from the answer RES(FstT(az,rg)) and the residual disturbance energy ClutterstT can be derived using equation (14) of [9]. The antenna patterns A S (α, β, fr) and A t(α, β, fr) in conjunction with the channel-independent geometry of the calibration acquisition give AstT(az,fr) the azimuth and range frequency dependent antenna gain relative to the reference target, as described in equation (6) of [9].

[0071] The strategy of the Fig. 8 is to use the available residual responses RES(FstT(az,rg)∀st∈C) to a single answer RES(FstT(az,rg)) with a significantly improved signal-to-noise ratio. As indicated by indices, the combined residual response RES(FstT(az,rg)) treated as if it were derived from a raw SAR data channel that uses a (virtual) antenna G for both transmission and reception. After providing the other input information required for optimization, namely the average residual interference energy ClutterGGT and the antenna diagrams (generally: A G (α, β, fr)), the original gaze direction estimation method according to [9] is applied to estimate the gaze direction correction parameters.

[0072] The fused signature for a respective reference target T (representing the majority of reference targets A, B, C) can then be determined as a weighted mean as follows: FGGT(az,rg)=1∑st∈CAstT(az)∑st∈CAstT(az)FstT(az,rg)

[0073] The antenna amplification integrated over the signal bandwidth is used: AstT(az)=∑fr|AstT(az,fr)|2.

[0074] The residual noise in the fused signature can then be determined as follows: ClutterGGT=max(〈ClutterstT〉st∈C−〈PggT(az,rg)−|FGGT(az,rg)|2〉az,rg,0), where <...> x the mean over x and PggT(az,rg)=1∑st∈CAstT(az)2∑st∈C|stT(az)FstT(az,rg)|2 denote the average residual energy.

[0075] The combined signature is associated with an effective antenna pattern according to AG(α,β,fr)=1|C|∑st∈C|As(α,β,fr)At(α,β,fr)| where |C| denotes the number of transmit and receive branches in C.

[0076] There is a certain degree of flexibility regarding the correction parameters (weights) used to determine the fused quantities. In the above formulation, greater emphasis is placed on signal components with a higher antenna gain in the interest of noise suppression. Other schemes are also conceivable. It should be noted that the individual responses FstT(az,rg) be added coherently to obtain the result FGGT(az,rg) The coherent superposition improves the signal-to-noise ratio and thus the accuracy of the direction estimation, combined with the following properties and advantages: - Additive coherent fusion reduces the effects of thermal noise on the SAR sensor instruments. - Fusion of signatures with antennas separated in a longitudinal direction reduces azimuth ambiguity. - Merging signatures with antennas separated cross-track reduces range ambiguity. - Fusion of signatures with different polarizations reduces the effects of thermal noise and reduces residual interference (including contributions from range and azimuth ambiguities).

[0077] In many cases, the set of antennas combined for improved joint direction estimation will include both longitudinal and transverse baselines and / or different polarizations, resulting in the suppression of both range and azimuth ambiguities as well as thermal noise and residual interference energy.

[0078] It is important to note that the coherent superposition used to obtain FGGT(az,rg) requires a certain phase calibration of all involved SAR raw data channels. Therefore, it is necessary to perform a calibration of the baseline and the relative phase positions prior to calibrating the antenna alignment. Such a high-precision calibration of the antenna baseline and phase positions is also described, for example, in [9]. References [1] Freeman, A. SAR calibration: An overview. IEEE Trans. Geosci. Remote Sens. 1992, 30, 1107-1121. [2] Werninghaus, R.; Buckreuss, S. The TerraSAR-X Mission and System Design. IEEE Trans. Geosci. Remote Sens. 2010, 48, 606-614. [3] Krieger, G.; Moreira, A.; Fiedler, H.; Hajnsek, I.; Werner, M.; Younis, M.; Zink, M. TanDEM-X: A Satellite Formation for High-Resolution SAR Interferometry. IEEE Trans. Geosci. Remote Sens. 2007, 45, 3317-3341. [4] Covello, F.; Battazza, F.; Coletta, A.; Lopinto, E.; Fiorentino, C.; Pietranera, L.; Valentini, G.; Zoffoli, S. COSMO-SkyMed an existing opportunity for observing the Earth. J. Geodyn. 2010; 49, 171-180. [5] Torres, R.; Snoeij, P.; Geudtner, D.; Bibby, D.; Davidson, M.; Attema, E.; Potin, P.; Rommen, B.; Floury, N.; Brown, M.; et al. GMES Sentinel-1 mission. Remote Sens. Environ. 2012, 120, 9-24. [6] Morena, L.C.; James, K.V.; Beck, J. An introduction to the RADARSAT-2 mission. Can. J. Remote Sens. 2004, 30, 221-234. [7] Reigber, A.; Scheiber, R.; Jager, M.; Prats-Iraola, P.; Hajnsek, I.; Jagdhuber, T.; Papathanassiou, K.P.; Nannini, M.; Aguilera, E.; Baumgartner, S.; et al. Very-High-Resolution Airborne Synthetic Aperture Radar Imaging: Signal Processing and Applications. Proc. IEEE 2013, 101, 759-783. [8] Reigber, A.; Jäger, M.; Fischer, J.; Horn, R.; Scheiber, R.; Prats, P.; Nottensteiner, A. System status and calibration of the F-SAR airborne SAR instrument. In Proceedings of the Geoscience and Remote Sensing Symposium (IGARSS), 2011 IEEE International, Vancouver, BC, Canada, 24-29 July 2011; pp. 1520-1523. [9] Jäger, M.; Scheiber, R.; Reigber, A. Robust, Model-Based External Calibration of Multi-Channel Airborne SAR Sensors Using Range Compressed Raw Data. Remote Sens. 2019, 11, 2674.

Claims

[1] Method for calibrating a SAR sensor with one or more receiving channels (CH1, CH2), each receiving channel (CH1, CH2) having a plurality (M, N) of receiving branches with an antenna (α 11 , ..., α 1N ; α 21 , ..., α 2M ) and with an analog-to-digital converter (ADC) for digitizing the signal received by the antenna (α 11 , ..., α 1N ; α 21 , ..., a 2M ) received SAR signal into SAR raw data (D 11 , ..., D 1N ; D 21 , ..., D 2M ) and a beamforming unit (DBF1, DBF2), wherein the beamforming unit (DBF1, DBF2) is configured to process the SAR raw data (D 11 , ..., D 1N ; D 21 , ..., D 2M ) of the plurality (M, N) of receiving branches with a weighting factor (α 1n (rg, αz), α 2m (rg, αz)), the weighted SAR raw data (D 11 , ..., D 1N ; D 21 , ..., D 2M) of the plurality (M, N) of reception branches and storing the SAR raw data obtained for a respective reception channel (CH1, CH2) in a data memory (SP1, SP2) for later processing, the method comprising the following steps: a) Adjusting the weight factors (α 1n (αz, rg), α 2m (az, rg)) of the receiving branches to obtain unweighted SAR raw data (D 11 , ..., D 1N ; D 21 , ..., D 2M ) to obtain; b) Performing a reference target analysis per receiving branch, in which the signatures of several reference targets detected by the SAR sensor in the unweighted SAR raw data (D 11 , ..., D 1N ; D 21 , ..., D 2M ) is compared with modeled signatures and a residual response of the signature is determined for each reference target in each raw data set; c) Determining a respective weight factor (α 1n (az, rg), α2m (az, rg)) for each of the plurality (M, N) of receiving branches based on the reference target analysis, wherein the determined weight factors (α 1n (az, rg), α 2m (αz, rg)) represent calibration parameters of the SAR sensor. [2] Method according to claim 1, characterized by that steps a) to c) are carried out, in particular once, before the SAR sensor is operated. [3] Method according to one of claims 1 or 2, characterized by that adjusting the weight factors (α 1n (αz, rg), α 2m (αz, rg)) of the receiving branches is carried out by a multiplexing operation by cyclically switching from receiving pulse to receiving pulse between the receiving branches or receiving branch groups, each receiving branch group comprising a combination of signals from several receiving branches. [4] Method according to claim 3, characterized by that the multiplex operation is carried out using binary weight factors (α 1n(az, rg), α 2m (αz, rg)) is implemented. [5] Method according to claim 3 or 4, characterized by that adjusting the weight factors (α 1n (az, rg), α 2m (αz, rg)) comprises a de-multiplexing step in which the receiving branches or receiving branch groups emerge from the de-multiplexing. [6] Method according to one of the preceding claims, characterized by that the SAR raw data (D 11 , ..., D 1N ; D 21 , ..., D 2M ) of the plurality (M, N) of reception branches or reception branch groups are subjected to a respective signal processing, wherein each reception branch group comprises a combination of signals from several reception branches. [7] Method according to claim 6, characterized by that the signal processing involves summing and shifting the center frequency of the raw SAR data (D 11 , ..., D 1N ; D 21 , ..., D 2M) of the plurality (M, N) of reception branches or reception branch groups. [8] Method according to claim 7, characterized by that adjusting the weight factors (α 1n (αz, rg), α 2m (αz, rg)) according to the following rule: αXn(az,rg)=exp(2πj(n−1)Naz). which are: α xn (αz, rg) is the weighting factor for the nth receiving branch or the nth receiving branch group, rg Integer index of the sampling point in the distance direction, starting at zero for the first sample in each received pulse, αz Integer azimuth pulse index in the calibration acquisition, starting at zero for the first transmit pulse of the data acquisition; N is the number of receiving branches or receiving branch groups. [9] Method according to one of claims 6 to 8, characterized by that adjusting the weight factors (α 1n (αz, rg), α 2m(az, rg)) comprises a step of undoing the phase shift, from which step the receiving branches or groups of receiving branches emerge. [10] Method according to one of claims 3 to 9, characterized by that the beam forming unit (DBF1, DBF2) is configured to calculate the weighting factors (α 1n (αz, rg), α 2m (αz, rg)) only in the azimuth direction in order to multiplex the plurality (M, N) of receive branches or receive branch groups to the output of the beamforming unit (DBF1, DBF2). [11] Method according to one of the preceding claims, characterized by that a determination of correction parameters for the viewing direction of several antennas of at least one group of antennas (u, v, w, x, y, z) of the SAR sensor is carried out. [12] Method according to claim 11, characterized bythat based on a reference target analysis in which the signatures of several reference targets (T ∈ A, B, C) in the SAR raw data channels (D 11 , ..., D 1N ; D 21 , ..., D 2M ) are compared with modeled signatures, for each reference target (T ∈ A, B, C) and each SAR raw data channel a residual response (FstT(az,rg)) the signatures are determined. [13] Method according to claim 12, characterized by that the residual responses (FstT(az,rg)) for each reference target (T ∈ A, B, C) in the SAR raw data channels of the group of antennas (u, v, w, x, y, z), resulting in a fused residual response for each reference target (T ∈ A, B, C) (FGGT(az,rg)) is obtained, from which the correction parameters for the viewing direction of several antennas of at least one group of antennas (u, v, w, x, y, z) of the SAR sensor are determined. [14] Method according to claim 13, characterized by that the fusion is an addition of the residual responses (FstT(az,rg)) for each reference target (T ∈ A, B, C) in the SAR raw data channels of the group of antennas (u, v, w, x, y, z). [15] Method according to claim 13 or 14, characterized by that the fusion is a weighted averaging of the residual responses (FstT(az,rg)) for each reference target (A, B, C) over the SAR raw data channels of the group of antennas (u, v, w, x, y, z), wherein an effective antenna pattern of the group of antennas (u, v, w, x, y, z) is processed as a weight. [16] Method according to one of claims 13 to 15, characterized by that the individual residual responses (FstT(az,rg)) be merged coherently. [17] Method according to one of claims 13 to 16, characterized by that the fused residual response (FGGT(az,rg)) and a fused antenna pattern of the group of antennas (u, v, w, x, y, z) are processed as input variables for determining the correction parameters. [18] A computer program product comprising instructions which, when the program is executed by a computer, cause the computer to carry out the method according to any one of claims 1 to 17. [19] Device for calibrating a SAR sensor with one or more receiving channels (CH1, CH2), each receiving channel (CH1, CH2) having a plurality (M, N) of receiving branches with an antenna (α 11 , ..., α 1N ; α 21 , ..., α 2M ) and with an analog-to-digital converter (ADC) for digitizing the signal received by the antenna (α 11 , ..., α 1N ; α 21 , ..., α 2M ) received SAR signal into SAR raw data (D 11 , ..., D 1N ; D 21 , ..., D 2M) and a beamforming unit (DBF1, DBF2), wherein the beamforming unit (DBF1, DBF2) is configured to process the SAR raw data (D 11 , ..., D 1N ; D 21 , ..., D 2M ) of the plurality (M, N) of receiving branches with a weighting factor (α 1n (rg, αz), α 2m (rg, αz)), the weighted SAR raw data (D 11 , ..., D 1N ; D 21 , ..., D 2M ) of the plurality (M, N) of reception branches and to store the SAR raw data obtained for a respective reception channel (CH1, CH2) in a data memory (SP1, SP2) for later processing, wherein the device comprises a processor which is configured to carry out the method according to one of claims 1 to 17.

Images