A multi-channel SAR full-scene data continuous imaging method

Abstract

The application provides a kind of multi-channel SAR full scene data continuous imaging method, after the scene of panorama data is divided, each single scene data obtained is respectively subjected to spectrum reconstruction, then according to the data after spectrum reconstruction, each scene is imaged, wherein, when each scene is imaged, the application considers the non-complete accumulation area of each half synthetic aperture before and after the scene is divided, part of data of each scene after spectrum reconstruction is directly used as the starting data of the next scene, so that the data amount corresponding to one synthetic aperture area does not need to be returned when the scene is divided each time, thereby the data operation amount of the next scene imaging can be reduced, and the full scene can be retained, so as to ensure the imaging quality.

Description

Technical Field

[0001] This invention belongs to the field of synthetic aperture radar signal processing technology, and particularly relates to a method for continuous imaging of multi-channel SAR full-scene data. Background Technology

[0002] Synthetic Aperture Radar (SAR) imaging technology is a crucial technology in remote sensing, playing a vital role in both military and civilian applications. High Resolution and Wide Swath (HRWS) is a major development direction for current SAR systems, and its most common implementation method is azimuth multichannel SAR. The echoes from multichannel systems are ambiguous in the azimuth direction, requiring signal reconstruction to obtain unambiguous echo signals before imaging.

[0003] In practical engineering, SAR has a long working time and collects a large amount of data at one time. Due to the limited processing and storage capabilities of hardware devices, it is generally necessary to process the data of the entire scene by scene. However, there are incomplete accumulation areas of half a synthetic aperture before and after the scene division. In order to ensure imaging quality, these areas should be removed after imaging. If the incomplete areas at the scene division are directly discarded, the scene will be lost. Therefore, a method that can be fast and continuously image the entire scene is needed. Summary of the Invention

[0004] To address the aforementioned issues, this invention provides a multi-channel SAR full-scene continuous imaging method, which features high efficiency and full-scene imaging capabilities.

[0005] A multi-channel SAR full-scene continuous imaging method involves dividing panoramic data into scenes, and then imaging each scene separately. Each scene consists of data from two or more sub-channels, with the total number of sub-channels denoted as N. c The imaging method for the first scene data is as follows:

[0006] Using the point N, which is the point of incomplete accumulation of unilateral azimuth in each sub-channel data, as the dividing point, the data of each sub-channel is divided into the front segment data and the back segment data.

[0007] All front-end data and back-end data are reconstructed by azimuth spectrum to obtain the unambiguous spectrum components corresponding to each front-end data and each back-end data. The unambiguous spectrum components corresponding to all front-end data are reconstructed into data A, and the unambiguous spectrum components corresponding to all back-end data are reconstructed into data B.

[0008] Perform an inverse orientation-to-Fourier transform on data A and data B to obtain data C and data D respectively;

[0009] Data C and data D are converted into single-channel echo signals and then imaged.

[0010] The imaging method for each scene after the first scene is as follows:

[0011] The reciprocal of the data D corresponding to the (M-1)th scene data ranked in the current Mth scene data (N·N) c The data is C corresponding to the Mth scene data, and the data D corresponding to the Mth scene data is obtained. Then, the single-channel echo signal equivalent to the data C and data D corresponding to the Mth scene data is used to image the Mth scene data.

[0012] Furthermore, the method for azimuth spectrum reconstruction of the front-end and back-end data is as follows:

[0013] The mapping relationship between the first sub-channel and subsequent sub-channels in the Doppler domain is obtained as follows:

[0014]

[0015] Where d is the spacing between the receiving antennas, v s Let m be the satellite velocity, and m be the sub-channel number, where m ∈ [1, N]. c ], f a f represents the azimuth frequency axis of a single sub-channel signal. a ∈[0,PRF], where PRF is the pulse repetition frequency. When reconstructing the azimuth spectrum of the preceding data, S1(f a ) represents the front-end data corresponding to the first sub-channel, S m (f a S1(f) represents the front-end data corresponding to the m-th sub-channel. When performing azimuth spectrum reconstruction on the back-end data, S1(f) a ) represents the back-end data corresponding to the first sub-channel, S m (f a ) represents the back-end data corresponding to the m-th sub-channel;

[0016] Acquire echo signals from each channel The mapping relationship between the unambiguous spectral components is as follows:

[0017]

[0018] Among them, f a,m For the azimuth frequency axis, and has P m Let m be the m-th unambiguous spectral component to be reconstructed;

[0019] The mapping relationship between the unambiguous spectral components and the unambiguous spectral components is represented in vector form as follows:

[0020]

[0021] Where X is the spatial steering matrix, expressed as:

[0022]

[0023] Through the inverse matrix X of the spatial guidance matrix X -1 Obtaining unambiguous spectral components

[0024]

[0025] Furthermore, the method for obtaining data D corresponding to the Mth scene is as follows:

[0026] Starting from the L×(M-1)+1th row of each sub-channel, take the following (L-2N) rows as the latter part of the data for each sub-channel corresponding to the Mth scene data, where L is the length of a single scene data;

[0027] The azimuth spectrum of each segment of data corresponding to the Mth scene is reconstructed, and the unambiguous spectral components corresponding to the Mth scene are extracted. The unambiguous spectral components corresponding to the Mth scene are then reconstructed into data B corresponding to the Mth scene.

[0028] After performing an inverse orientation-to-Fourier transform on the data B corresponding to the Mth scene, we obtain the data D corresponding to the Mth scene.

[0029] Furthermore, after imaging the single-channel echo signals corresponding to each scene, the incompletely accumulated portions in the 2N rows on both sides of the azimuth direction of the imaging results are removed, leaving the fully accumulated portions as the final imaging scene.

[0030] Furthermore, based on the spatial sampling positions of each sub-channel, data C and data D are equivalent to single-channel echo signals.

[0031] Beneficial effects:

[0032] 1. This invention provides a multi-channel SAR full-scene data continuous imaging method. After dividing the panoramic data into scenes, the spectrum of each obtained single scene data is reconstructed. Then, scene-by-scene imaging is performed based on the spectrum-reconstructed data. In scene-by-scene imaging, this invention takes into account the incomplete accumulation area of ​​half of the synthetic aperture before and after the scene division point. After spectrum reconstruction, a portion of the data of each scene is retained and directly used as the starting data of the next scene. This means that the data does not need to be backed up by the amount of data corresponding to a synthetic aperture area each time the scene is divided, thereby reducing the amount of data computation for the next scene imaging while retaining the full scene and ensuring imaging quality.

[0033] 2. This invention provides a multi-channel SAR full-scene continuous imaging method. By using an inverse filter algorithm, unambiguous spectral components are extracted from the front and back segments of the scene data, which can solve the problems of non-uniform azimuth sampling and spectral aliasing of a single sub-channel caused by multi-channel systems.

[0034] 3. This invention provides a multi-channel SAR full-scene data continuous imaging method. After imaging the single-channel echo signal corresponding to each scene data, the incompletely accumulated part is removed, and the fully accumulated part is left as the final imaging scene, which can ensure the imaging quality. Attached Figure Description

[0035] Figure 1 This is an overall flowchart of a multi-channel SAR full-scene data continuous imaging method according to the present invention;

[0036] Figure 2 This is the image result of the first scene;

[0037] Figure 3 This is the image result for the second scene;

[0038] Figure 4 This is the full-scene imaging result. Detailed Implementation

[0039] To enable those skilled in the art to better understand the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings.

[0040] Example 1

[0041] In practical engineering, the full-scene echo data is very large, requiring scene segmentation. If the incomplete accumulation regions of half the synthetic aperture before and after the scene segmentation point are not considered before imaging, the incomplete accumulation regions on both sides in the azimuth direction are removed after imaging to ensure image quality, resulting in scene loss. If the incomplete accumulation regions of half the synthetic aperture before and after the scene segmentation point are considered before imaging, the data needs to be backed up by the amount of data corresponding to one synthetic aperture region each time the scene is segmented, resulting in low work efficiency. This invention retains a portion of the data after spectrum reconstruction as the starting data for the next scene, which can reduce the amount of computation for the next scene data while preserving the entire scene, as detailed below:

[0042] A multi-channel SAR full-scene continuous imaging method involves dividing panoramic data into scenes, and then imaging each scene separately. Each scene consists of data from two or more sub-channels, with the total number of channels denoted as N. cIt should be noted that multi-channel panoramic data is composed of panoramic data from each sub-channel. Therefore, each scene data in a sub-channel is used as sub-channel data. In other words, the size of the sub-channel data defined in this invention is only the size of a single scene data in a single channel.

[0043] The imaging method for the first scene data is as follows:

[0044] Using the point N, which is an incomplete accumulation point in the unilateral azimuth of each sub-channel data, as the dividing point, the data of each sub-channel is divided into the front segment data and the back segment data respectively;

[0045] All front-end and back-end data are reconstructed by azimuth spectrum, and the unambiguous spectrum components corresponding to each front-end data and each back-end data are extracted. The ambiguous spectrum components corresponding to all front-end data are reconstructed into data A, and the unambiguous spectrum components corresponding to all back-end data are reconstructed into data B.

[0046] After performing orientation-to-Fourier inverse transforms on data A and data B respectively, we obtain data C and data D.

[0047] After converting data C and data D into equivalent single-channel echo signals based on the spatial sampling positions of each sub-channel, imaging is performed based on the single-channel echo signals. The imaging result at this point contains incompletely accumulated portions; after removing the N·N values ​​on both sides of the azimuth axis in the imaging result... c The incompletely accumulated portion of the image is used to determine the final image scene, leaving the fully accumulated portion.

[0048] The imaging method for each scene after the first scene is as follows:

[0049] The reciprocal of the data D corresponding to the (M-1)th scene data ranked in the current Mth scene data (N·N) c The data C corresponding to the Mth scene is used as the data line, and the data D corresponding to the Mth scene is obtained. Then, the single-channel echo signal equivalent to the data C and data D corresponding to the Mth scene is used to image the Mth scene. Similarly, the imaging result at this time also contains the incomplete accumulation part. The N·N values ​​on both sides of the azimuth direction in the imaging result are removed. C The incompletely accumulated portion is used to determine the final image scene, leaving the fully accumulated portion.

[0050] The method for obtaining data D corresponding to the Mth scene is as follows:

[0051] Starting from the L×(M-1)+1th row of each sub-channel, take the following (L-2N) rows as the latter part of the data for each sub-channel corresponding to the Mth scene data, where L is the length of the single scene data of each sub-channel;

[0052] The azimuth spectrum of each segment of data corresponding to the Mth scene is reconstructed, and the unambiguous spectral components corresponding to the Mth scene are extracted. The unambiguous spectral components corresponding to the Mth scene are then reconstructed into data B corresponding to the Mth scene.

[0053] After performing an inverse orientation-to-Fourier transform on the data B corresponding to the Mth scene, we obtain the data D corresponding to the Mth scene.

[0054] Furthermore, the method for azimuth spectrum reconstruction of the front-end and back-end data is as follows:

[0055] The mapping relationship between the first sub-channel and subsequent sub-channels in the Doppler domain is obtained as follows:

[0056]

[0057] Where d is the spacing between the receiving antennas, v s Let m be the satellite velocity, and m be the sub-channel number, where m ∈ [1, N]. c ], f a f represents the azimuth frequency axis of a single sub-channel signal. a ∈[0,PRF], where PRF is the pulse repetition frequency. When reconstructing the azimuth spectrum from the preceding data, S1(f a ) represents the front-end data corresponding to the first sub-channel, S m (f a S1(f) represents the front-end data corresponding to the m-th sub-channel. When performing azimuth spectrum reconstruction on the back-end data, S1(f) a ) represents the back-end data corresponding to the first sub-channel, S m (f a ) represents the back-end data corresponding to the m-th sub-channel;

[0058] Acquire echo signals from each channel The mapping relationship between the unambiguous spectral components is as follows:

[0059]

[0060] For example, the mapping relationship between the two endpoints and the unambiguous spectral components is as follows:

[0061]

[0062] Among them, f a,m For the azimuth frequency axis, and has P m Let m be the m-th unambiguous spectral component to be reconstructed;

[0063] The mapping relationship between the unambiguous spectral components and the unambiguous spectral components is represented in vector form as follows:

[0064]

[0065] Where X is the spatial steering matrix, expressed as:

[0066]

[0067] Through the inverse matrix X of the spatial guidance matrix X -1 Obtaining unambiguous spectral components

[0068]

[0069] Example 2

[0070] like Figure 1 As shown, the following description uses a dual-channel example to further illustrate a multi-channel SAR full-scene data continuous imaging method of the present invention.

[0071] S1: Data Segmentation. Taking a dual-channel setup as an example, assuming the data volume of one scene in a sub-channel is 16k rows, and if the incomplete accumulation point in the azimuth direction of a single sub-channel is N, the data of a single sub-channel starts from the first pulse repetition time PRT. The 16k data is divided into two parts: the first 2N rows and the last (16k-2N) rows. The dual-channel setup will then obtain four parts of data: 2N rows, 2N rows, (16k-2N) rows, and (16k-2N) rows. These are denoted as A1 for the first sub-channel, A2 for the second sub-channel, B1 for the last sub-channel, and B2 for the second sub-channel.

[0072] S2: Azimuth spectrum reconstruction, which uses an inverse filter algorithm to solve the problems of non-uniform azimuth sampling and aliasing of individual sub-channel spectra caused by multi-channel systems. Details are as follows:

[0073] (1) The spectrum of the two parts, A1 and A2, with 2N rows, is reconstructed to obtain 4N rows of unambiguous spectral component data, denoted as A.

[0074] (2) The spectrum reconstruction of the two parts of data B1 (16k-2N) and data B2 (16k-2N) results in 2*(16k-2N) rows of unambiguous spectrum component data, denoted as B.

[0075] Assuming the spacing between the receiving antennas is d, then the spacing between the equivalent phase centers is... The satellite's speed is v s The signal S of the m-th channel m This is equivalent to the delay of the first channel. The mapping relationship between the first and second sub-channels in the Doppler domain is as follows:

[0076]

[0077] S1(f a If the spectrum of data A1 and data A2 is reconstructed at this time, S1(f a ) represents data A1, S2(f a Let S1(f) be data A2; if we then perform spectrum reconstruction on data B1 and data B2, S1(f) a ) represents data B1, S2(f a () is data B2;

[0078] In a multi-channel system, the azimuth direction of a single sub-channel is undersampled, meaning the pulse repetition frequency (PRF) is less than the Doppler bandwidth. Taking a dual-channel system as an example, the signal S of each channel... m (f a ) and each unambiguous spectral component P i The relationship between them is:

[0079]

[0080] Among them, f a,1 ∈[-PRF,0],f a,2 ∈[0,PRF].

[0081] The above formula can be expressed in vector form as follows:

[0082] [S1(f a S2(f) a )]=[P1 P2]X

[0083] Where X is the spatial steering matrix, expressed as:

[0084]

[0085] Then there is no ambiguity spectral component P i for:

[0086] [P1 P2]=[S1(f a S2(f) a )]X -1

[0087] In other words, when performing spectrum reconstruction on dual channels, data A1 and data A2 are first used as S1(f a ) and S2(f a Extract P1 and P2, and then reconstruct the data A from P1 and P2; then use data B1 and data B2 as S1(f) respectively. a ) and S2(fa Extract data P1 and P2, and then reconstruct the data B from P1 and P2.

[0088] S3: After spectrum reconstruction, the uniformly sampled signal is equivalent to a single-channel echo signal. The specific steps are as follows:

[0089] (1) The data A in step S2 is transformed into a two-dimensional time domain signal by azimuth-to-Fourier inverse transformation, which yields the data C.

[0090] (2) The data B in step S2 is transformed into a two-dimensional time domain signal through azimuth-to-Fourier inverse transformation, which yields the data D.

[0091] (3) Based on the spatial sampling position between each channel, data C and data D are equivalent to single-channel 32k echo signals.

[0092] (4) Take 4N rows from bottom to top in the data D and use them as the first 4N rows of the next scene, which is the C of the next scene.

[0093] S4: Imaging and post-imaging processing.

[0094] (1) The 32k data obtained in step S3(3) is processed by an imaging algorithm.

[0095] (2) Remove the incomplete accumulation portion with an azimuth of 2N on each side of the imaging result obtained in step S4(1), leaving the final imaging scene with a complete accumulation size of (32k-4N).

[0096] S5: Starting from the second scene, and taking the current scene as the Mth scene as an example, take the next (16k-2N) rows from the 16K×(M-1)+1th row of each sub-channel as the last part of the data of the Mth scene, that is, get the data B1 and data B2 in step S1. The newly acquired data B1 and data B2 are re-operated in step S2(2) to get the data B corresponding to the Mth scene data. The newly acquired data B is re-operated in step S3(2) to get the data D corresponding to the Mth scene. At this time, the data D is combined with the data C in step S3(4) to perform the operations of step S3(3) and step S4 until all data is processed.

[0097] like Figures 2-3 As shown, the images are the first scene data, the second scene data, and the imaging results based on the first scene data, the second scene data, and the panoramic data, respectively. It can be seen that the present invention retains some data after spectrum reconstruction as the starting data for the next scene, which can obtain good imaging results.

[0098] Of course, the present invention may have other various embodiments. Without departing from the spirit and essence of the present invention, those skilled in the art can make various corresponding changes and modifications according to the present invention, but these corresponding changes and modifications should all fall within the protection scope of the appended claims.

Claims

1. A method for continuous imaging of multi-channel SAR full-scene data, characterized in that, After segmenting the panoramic data into scenes, each scene is then imaged separately. Each scene consists of data from two or more sub-channels, and the total number of sub-channels is denoted as [missing information]. The imaging method for the first scene data is as follows: Using the point N, which is the point of incomplete accumulation of unilateral azimuth in each sub-channel data, as the dividing point, the data of each sub-channel is divided into the front segment data and the back segment data. All front-end data and back-end data are reconstructed by azimuth spectrum to obtain the unambiguous spectrum components corresponding to each front-end data and each back-end data. The unambiguous spectrum components corresponding to all front-end data are reconstructed into data A, and the unambiguous spectrum components corresponding to all back-end data are reconstructed into data B. Perform an inverse orientation-to-Fourier transform on data A and data B to obtain data C and data D respectively; Data C and data D are converted into single-channel echo signals and then imaged. The imaging method for each scene after the first scene is as follows: The reciprocal of the data D corresponding to the (M-1)th scene data ranked in the current Mth scene data. The data is C corresponding to the Mth scene data, and the data D corresponding to the Mth scene data is obtained. Then, the single-channel echo signal equivalent to the data C and data D corresponding to the Mth scene data is used to image the Mth scene data.

2. The multi-channel SAR full-scene continuous imaging method as described in claim 1, characterized in that, The method for reconstructing the azimuth spectrum from the front-end and back-end data is as follows: The mapping relationship between the first sub-channel and subsequent sub-channels in the Doppler domain is obtained as follows: in, The spacing between receiving antennas, For satellite speed, This is the sequence number of the sub-channel. This represents the azimuth frequency axis of a single sub-channel signal. , The pulse repetition frequency is used when reconstructing the azimuth spectrum from the preceding data. This represents the front-end data corresponding to the first sub-channel. Indicates the first When reconstructing the azimuth spectrum of the data corresponding to each sub-channel, the preceding data is used... This indicates the subsequent data corresponding to the first sub-channel. Indicates the first The back-end data corresponding to each sub-channel; Acquire echo signals from each channel ~ The mapping relationship between the unambiguous spectral components is as follows: in, For the azimuth frequency axis, and has , For each One unambiguous spectral component to be reconstructed; ~ The mapping relationship between the unambiguous spectral components and the unambiguous spectral components is represented in vector form as follows: in, The spatial guidance matrix is ​​expressed as: Through the spatial guidance matrix inverse matrix Obtaining unambiguous spectral components ~ : 。 3. The multi-channel SAR full-scene continuous imaging method as described in claim 1, characterized in that, The method for obtaining data D corresponding to the Mth scene is as follows: Starting from the L×(M-1)+1th row of each sub-channel, take the following (L-2N) rows as the latter part of the data for each sub-channel corresponding to the Mth scene data, where L is the length of a single scene data; The azimuth spectrum of each segment of data corresponding to the Mth scene is reconstructed, and the unambiguous spectral components corresponding to the Mth scene are extracted. The unambiguous spectral components corresponding to the Mth scene are then reconstructed into data B corresponding to the Mth scene. After performing an inverse orientation-to-Fourier transform on the data B corresponding to the Mth scene, we obtain the data D corresponding to the Mth scene.

4. A multi-channel SAR full-scene continuous imaging method as described in any one of claims 1 to 3, characterized in that, After imaging the single-channel echo signals corresponding to each scene, the azimuth signals on both sides of the imaging result are removed. The incompletely accumulated portion of the image is used to determine the final image scene, leaving the fully accumulated portion.

5. A multi-channel SAR full-scene continuous imaging method as described in any one of claims 1 to 3, characterized in that, Based on the spatial sampling position of each sub-channel, data C and data D are equivalent to single-channel echo signals.

Images