Building Area Extraction Method for π / 4 Compact Polarimetric Synthetic Aperture Radar

Abstract

The present invention discloses a π / 4 reduced polarization synthetic aperture radar building region extraction method, which utilizes reduced polarization features to reasonably extract building regions, and provides a new idea for the π / 4 reduced polarization SAR building extraction method. The method comprises: π / 4 reduced polarization SAR image data preprocessing; obtaining the preprocessed π / 4 reduced polarization SAR image, and calculating the reduced polarization features; utilizing the reduced polarization features, calculating a water body extractor, an orthogonal building region extractor where the building region is parallel / perpendicular to the radar line of sight direction, and an inclined building region extractor where the building region has a large inclination angle with the radar line of sight direction; obtaining thresholds of a water body extractor, an orthogonal building extractor, and an inclined building region extractor by selecting typical regions for statistical analysis, and using the thresholds to extract the building region, and obtaining a building region extraction result.

Description

Technical Field

[0001] The invention relates to the field of reduced polarimetric synthetic aperture radar remote sensing, and in particular to a π / 4 reduced polarimetric synthetic aperture radar (SAR) building area extraction method. Background Art

[0002] With the rapid development of synthetic aperture radar (SAR) measurement technology, the research direction has evolved towards multi-polarization, multi-frequency, and multi-angle. Among them, polarimetric SAR, as an important branch of SAR measurement technology, plays an important role in ground object classification, target recognition, and disaster detection. As an emerging technology of polarimetric SAR, reduced polarimetric SAR has a large bandwidth. Currently, there are few application methods based on reduced polarimetric SAR, and it has great application potential in various application fields.

[0003] As one of the important types of land features, the density and quantity of buildings can be used as the basis for applications such as urban planning, population density estimation and disaster assessment. The accuracy of building area extraction is very important for application evaluation. At present, the existing building extraction methods mainly include two categories: building area extraction methods based on scattering models and building area extraction methods based on polarization parameters. The building area extraction method based on scattering models selects points with dominant dihedral scattering as building area candidate points to distinguish between building areas and non-building areas. This method is an unsupervised binary classification method, and the building extraction result strongly depends on the accuracy of the decomposition result; the building area extraction method based on polarization parameters selects the threshold of polarization parameters through statistical analysis to distinguish between building areas and non-building areas. This method is a supervised binary classification method, and the building extraction result strongly depends on the robustness of polarization parameters. Limited by the number of options and application limitations of existing methods, the effectiveness of the method is still an open problem. How to use emerging technologies for building area extraction, enrich building extraction methods, and reasonably and correctly extract building areas is still a technical challenge. Summary of the invention

[0004] In view of this, the present invention proposes a π / 4 reduced polarimetric SAR building area extraction method, the method comprising:

[0005] S1: π / 4 reduced polarimetric SAR image data preprocessing;

[0006] S2: Obtain the preprocessed π / 4 reduced polarimetric SAR image and calculate the reduced polarimetric features;

[0007] S3: Using the reduced polarization feature, a water body extractor, an orthogonal building area extractor where the building area is parallel / perpendicular to the radar line of sight, and an inclined building area extractor where the building area has a large inclination angle with the radar line of sight are calculated;

[0008] S4: By selecting typical areas for statistical analysis, thresholds of water body extractor, orthogonal building extractor and inclined building area extractor are obtained, and the thresholds are used to extract building areas to obtain building area extraction results.

[0009] In the above scheme, the method further includes:

[0010] After completing preprocessing operations such as filtering and multi-viewing on the full polarimetric SAR image, the full polarimetric SAR data is converted into π / 4 reduced polarimetric SAR data by using the conversion relationship between the full polarimetric SAR data and the reduced polarimetric SAR data for subsequent processing. The specific implementation method of this step is as follows:

[0011] Obtain the ensemble average <[T]> of the coherence matrix of the fully polarized SAR image:

[0012]

[0013] Among them, T ij (i=1,2,3; j=1,2,3) represents the i-th row and j-th column element of the ensemble average value of the full polarization coherence matrix <[T]>; the superscript “H” represents the conjugate transpose of the matrix; k is the Pauli basis,

[0014] Among them, S HH is the backscatter of vertical transmission and vertical reception, S HV is the backscatter with vertical transmission and horizontal reception, S VV It is the backscatter with horizontal transmission and horizontal reception.

[0015] In the π / 4 reduced polarimetric SAR, the 2×2 coherence matrix can be expressed by the Stokes vector as:

[0016]

[0017] Among them, the parameter J ij (i=1,2;j=1,2) represents the i-th row and j-th column element of the π / 4 reduced polarimetric SAR coherence matrix, G=(g0,g1,g2,g3) T is the Stokes vector, g k (k=0,1,2,3) represents the element of the kth row of the Stokes vector. The corresponding relationship between the Stokes vector and the fully polarized SAR data is:

[0018]

[0019] In the above scheme, the method further comprises the following sub-steps:

[0020] S21: Using the reduced polarimetric SAR coherence matrix J2 obtained in step S1, non-negative real eigenvalues ​​λ1 and λ2 are calculated. The expressions of λ1 and λ2 are:

[0021]

[0022]

[0023] S22: Calculate the reduced polarimetric radar vegetation index (RVI) and anisotropy (PA), construct RVI and PA parameters, and construct the RVI / PA plane for classifying different types of land objects. The expressions of the above two reduced polarimetric parameters are:

[0024]

[0025] Among them, the parameter μ=λ2 / λ1.

[0026] S23: Calculates the reduced polarimetric parameters (P) that can be used to describe areas of inclined buildings that have large inclination angles to the radar line of sight. LB ), the expression of this parameter is:

[0027]

[0028] In the above scheme, the method further includes:

[0029] Using RVI, PA and P LB , construct a water body extractor, an orthogonal building area extractor where the building area is parallel / perpendicular to the radar line of sight, and an inclined building area extractor where the building area has a large inclination angle with the radar line of sight. The expressions of the above extractors are:

[0030] Water Extractor: D W =(1-RVI)·PA

[0031] Orthogonal Building Area Extractor: D OB =(1-RVI)·(1-PA)

[0032] Sloped Building Area Extractor:

[0033] In the above scheme, the method further comprises the following sub-steps:

[0034] S41: Select typical areas such as orthogonal building areas, inclined building areas and non-building areas in the image for statistical analysis, draw a statistical histogram, select the intersection of the histogram curve as the judgment threshold of different types of areas, and intuitively compare and analyze the thresholds corresponding to different extractors;

[0035] S42: Using the extractor threshold obtained in step S41, the extraction results of the inclined building area and the orthogonal building area are obtained respectively:

[0036] Orthogonal building area extraction results:

[0037] Sloped building area extraction results:

[0038] Among them, T W , T OB and T LB These correspond to the thresholds for the water body extractor, the orthogonal building area extractor, and the inclined building extractor, respectively.

[0039] S43: Merge the inclined building area extraction result and the orthogonal building area extraction result obtained in step S42 to obtain the final research area building extraction result.

[0040] The advantages and beneficial effects of the present invention are as follows: by constructing a building area extractor with different inclination angles to the radar line of sight, the building area of ​​the study area can be extracted more reasonably and accurately, which provides a new idea for the reduced polarimetric SAR building area extraction technology and improves the current situation where the existing reduced polarimetric SAR building extraction methods are relatively scarce. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] The drawings illustrate generally, by way of example and not limitation, various embodiments discussed herein.

[0042] Figure 1 A schematic flow chart of a method for extracting building areas using a π / 4 reduced polarimetric SAR according to an embodiment of the present invention;

[0043] Figure 2 Typical areas such as orthogonal building areas, inclined building areas and non-building areas selected for determining extractor thresholds in the embodiments of the present invention;

[0044] Figure 3 This is the building area extraction result of the application scenario of the embodiment of the present invention. DETAILED DESCRIPTION

[0045] In order to enable a more detailed understanding of the features and technical contents of the embodiments of the present invention, the implementation of the embodiments of the present invention is described in detail below in conjunction with the accompanying drawings. The attached drawings are for reference only and are not intended to limit the embodiments of the present invention.

[0046] like Figure 1 As shown, the π / 4 reduced polarimetric SAR building area extraction method according to the embodiment of the present invention comprises the following steps:

[0047] S1: π / 4 reduced polarimetric SAR image data preprocessing;

[0048] S2: Obtain the preprocessed π / 4 reduced polarimetric SAR image and calculate the reduced polarimetric features;

[0049] S3: Using the reduced polarization feature, a water body extractor, an orthogonal building area extractor where the building area is parallel / perpendicular to the radar line of sight, and an inclined building area extractor where the building area has a large inclination angle with the radar line of sight are calculated;

[0050] S4: By selecting typical areas for statistical analysis, thresholds of water body extractor, orthogonal building extractor and inclined building area extractor are obtained, and the thresholds are used to extract building areas to obtain building area extraction results.

[0051] Taking the fully polarimetric SAR image data acquired by GF-3 in the San Francisco area of ​​the United States as an example, the effectiveness of the present invention in practical engineering applications is verified. The image size is 7681×5833 pixels (azimuth×range), and the resolutions in the range and azimuth directions are both 8m. Before subsequent processing, a filter with a window size of 7×7 is used for filtering.

[0052] Specifically, the step S1 further includes:

[0053] After completing the pre-processing operations such as filtering and multi-viewing on the full polarimetric SAR image acquired by GF-3, the full polarimetric SAR data is converted into π / 4 reduced polarimetric SAR data by using the conversion relationship between the full polarimetric SAR data and the reduced polarimetric SAR data for subsequent processing. The specific implementation method of this step is as follows:

[0054] Obtain the ensemble average <[T]> of the coherence matrix of the fully polarized SAR image:

[0055]

[0056] Among them, T ij (i=1,2,3; j=1,2,3) represents the i-th row and j-th column element of the ensemble average value of the full polarization coherence matrix <[T]>; the superscript “ H" represents the conjugate transpose of the matrix; k is the Pauli basis,

[0057] Among them, S HH is the backscatter of vertical transmission and vertical reception, S HV is the backscatter with vertical transmission and horizontal reception, S VV It is the backscatter with horizontal transmission and horizontal reception.

[0058] In the π / 4 reduced polarimetric SAR, the 2×2 coherence matrix can be expressed by the Stokes vector as:

[0059]

[0060] Among them, the parameter J ij (i=1,2;j=1,2) represents the i-th row and j-th column element of the π / 4 reduced polarimetric SAR coherence matrix, G=(g0,g1,g2,g3) T is the Stokes vector, g k (k=0,1,2,3) represents the element of the kth row of the Stokes vector. The corresponding relationship between the Stokes vector and the fully polarized SAR data is:

[0061]

[0062] Specifically, the step S2 further includes the following sub-steps:

[0063] S21: Using the reduced polarimetric SAR coherence matrix J2 obtained in step S1, non-negative real eigenvalues ​​λ1 and λ2 are calculated. The expressions of λ1 and λ2 are:

[0064]

[0065]

[0066] S22: Calculate the reduced polarimetric radar vegetation index (RVI) and anisotropy (PA), construct RVI and PA parameters, and construct the RVI / PA plane for classifying different types of land objects. The expressions of the above two reduced polarimetric parameters are:

[0067]

[0068] Among them, the parameter μ=λ2 / λ1.

[0069] S23: Calculates the reduced polarimetric parameters (P) that can be used to describe areas of inclined buildings that have large inclination angles to the radar line of sight. LB ), the expression of this parameter is:

[0070]

[0071] Specifically, step S3 includes:

[0072] Using RVI, PA and P LB , construct a water body extractor, an orthogonal building area extractor where the building area is parallel / perpendicular to the radar line of sight, and an inclined building area extractor where the building area has a large inclination angle with the radar line of sight. The expressions of the above extractors are:

[0073] Water Extractor: D W =(1-RVI)·PA

[0074] Orthogonal Building Area Extractor: D OB =(1-RVI)·(1-PA)

[0075] Sloped Building Area Extractor:

[0076] Specifically, the step S4 further includes the following sub-steps:

[0077] S41: Select typical areas such as orthogonal building areas, inclined building areas and non-building areas in the image for statistical analysis. The selected typical areas are as follows: Figure 2 Draw a statistical histogram, select the intersection of the histogram curve as the threshold for different types of regions, and visually compare and analyze the thresholds corresponding to different extractors;

[0078] S42: Using the extractor threshold obtained in step S41, the extraction results of the inclined building area and the orthogonal building area are obtained respectively:

[0079] Orthogonal building area extraction results:

[0080] Sloped building area extraction results:

[0081] Among them, T W , T OB and T LB These correspond to the thresholds for the water body extractor, the orthogonal building area extractor, and the inclined building extractor, respectively.

[0082] S43: The inclined building area extraction result obtained in step S42 is combined with the orthogonal building area extraction result to obtain the final building area extraction result of the research area. The final building area extraction result of the building extraction method proposed in the present invention is as follows: Figure 3 shown.

[0083] So far, the present embodiment has been described in detail with reference to the accompanying drawings. Based on the above description, those skilled in the art should have a clear understanding of the π / 4 reduced polarimetric SAR building region extraction method of the present invention.

[0084] The methods provided by the present invention can be arbitrarily combined without conflict to obtain new method embodiments.

[0085] The above is only a specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art can easily think of changes or substitutions within the technical scope disclosed by the present invention, which should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention should be based on the protection scope of the claims.

Claims

1. A method for extracting building areas using π / 4 reduced polarization synthetic aperture radar, characterized in that: The method comprises: S1: π / 4 reduced polarimetric SAR image data preprocessing; S2: Obtain the preprocessed π / 4 reduced polarimetric SAR image and calculate the reduced polarimetric features; S3: Using the reduced polarization feature, a water body extractor, an orthogonal building area extractor where the building area is parallel / perpendicular to the radar line of sight, and an inclined building area extractor where the building area has a large inclination angle with the radar line of sight are calculated; S4: by selecting typical areas for statistical analysis, thresholds of water body extractor, orthogonal building extractor and inclined building area extractor are obtained, and the thresholds are used to extract building areas to obtain building area extraction results; The step S1 comprises: After filtering and multi-view preprocessing operations are completed on the full polarimetric SAR image, the full polarimetric SAR data is converted into π / 4 reduced polarimetric SAR data by using the conversion relationship between the full polarimetric SAR data and the reduced polarimetric SAR data for subsequent processing. The specific implementation method of this step is as follows: Obtain the ensemble average <[T]> of the coherence matrix of the fully polarized SAR image: Among them, T ij (i=1,2,3; j=1,2,3) represents the i-th row and j-th column element of the ensemble average value of the full polarization coherence matrix <[T]>; the superscript "H" represents the conjugate transpose of the matrix; k is the Pauli basis, Among them, S HH is the backscatter of vertical transmission and vertical reception, S HV is the backscatter with vertical transmission and horizontal reception, S VV Backscatter for horizontal transmission and horizontal reception; The step S2 comprises the following sub-steps: S21: Using the reduced polarimetric SAR coherence matrix J2 obtained in step S1, non-negative real eigenvalues ​​λ1 and λ2 are calculated. The expressions of λ1 and λ2 are: S22: Calculate the reduced polarimetric radar vegetation index RVI and the anisotropy PA, construct RVI and PA parameters, and construct an RVI / PA plane for classifying different types of land objects. The expressions of the reduced polarimetric radar vegetation index RVI and the anisotropy PA are as follows: Among them, the parameter μ = λ2 / λ1; S23: Calculate the reduced polarimetric parameter P that can be used to describe areas of inclined buildings with large inclination angles to the radar line of sight LB , the expression of this parameter is: The step S3 comprises: Using RVI, PA and P LB , construct a water body extractor, an orthogonal building area extractor where the building area is parallel / perpendicular to the radar line of sight, and an inclined building area extractor where the building area has a large inclination angle with the radar line of sight. The expressions of the above extractors are: Water Extractor: D W =(1-RVI)·PA Orthogonal Building Area Extractor: D OB =(1-RVI)·(1-PA) Sloped Building Area Extractor: The step S4 comprises the following sub-steps: S41: selecting typical areas of orthogonal building areas, inclined building areas and non-building areas in the image for statistical analysis, drawing statistical histograms, selecting the intersection points of the histogram curves as the judgment thresholds for different types of areas, and visually comparing and analyzing the thresholds corresponding to different extractors; S42: Using the extractor threshold obtained in step S41, the extraction results of the inclined building area and the orthogonal building area are obtained respectively: Orthogonal building area extraction results: Sloped building area extraction results: Among them, T W , T OB and T LB corresponding to the thresholds of water body extractor, orthogonal building area extractor and inclined building extractor, respectively; S43: Merge the inclined building area extraction result and the orthogonal building area extraction result obtained in step S42 to obtain the final research area building extraction result.

2. The π / 4 reduced polarization synthetic aperture radar building area extraction method according to claim 1, characterized in that: The step S1 comprises: In the π / 4 reduced polarimetric SAR, the 2×2 coherence matrix is ​​expressed by the Stokes vector as: Among them, the parameter J ij (i=1,2;j=1,2) represents the i-th row and j-th column element of the π / 4 reduced polarimetric SAR coherence matrix, G=(g0,g1,g2,g3) Τ is the Stokes vector, g k (k=0,1,2,3) represents the element of the kth row of the Stokes vector; the corresponding relationship between the Stokes vector and the fully polarized SAR data is:

Images