High-definition imaging apparatus and method

• US 20030071750A1
• Filed: 02/28/2002
• Published: 04/17/2003
• Est. Priority Date: 03/02/2001
• Status: Active Grant

• Current Assignee: Massachusetts Institute of Technology
• Original Assignee: Massachusetts Institute of Technology
• Inventors: Benitz, Gerald R.
• Granted Patent: US 6,608,585 B2
• Time in Patent Office: Days
• US Class Current: 342/25
• CPC Class Codes: G01S 13/9011 with frequency domain processing of the SAR signals in azimuth G01S13/9005 takes precedence

Abstract

A high-definition radar imaging system and method receives image data and adaptively processes the image the data to provide a high resolution image. The imaging technique employs adaptive processing using a constrained minimum variance method to iteratively compute the high-definition image. The high-definition image I is expressed in range and cross-range as I(r,c)=minω^HRω, where ω is a weighting vector and R is a covariance matrix of the image data. A solution for I(r,c) is approximated by i) forming Y=[x₁ . . . x_K]^T/{square root}{square root over (K)} where x₁ . . . x_k are beamspace looks formed from image domain looks and with y₁, y₂, and y₃ denoting the K×1 columns of Y; ii) computing r₂₁=y₂^Ty₁ and r₃₁=y₃^Ty₁, and b=r₂₁y₂+r₃₁y₃; computing γ as $\gamma =\min \left(\frac{r_{21}^2+r_{31}^2}{b^Tb},\sqrt{\frac{\beta -1}{r_{21}^2+r_{31}^2}}\right);$

and iii) computing I(r,c) as I(r,c)=∥y₁−γb∥².

Claims

1. A method of processing image data to produce a high-definition image, comprising the steps of:

receiving the image data; and

adaptively processing the image data using a constrained minimum variance method to iteratively compute the high-definition image, wherein the high-definition image I is expressed in range and cross-range as I(r,c)=minω

^HRω

, where ω

is a weighting vector and R is a covariance matrix of the image data, wherein a solution for I(r,c) is approximated by i) forming Y=[x₁ . . . x_K]^T/{square root}{square root over (K)}, where x₁ . . . x_k are beamspace looks formed from image domain looks and with y₁, y₂, and y₃ denoting the K×

1 columns of Y;

ii) computing r₂₁=y₂^Ty₁ and r₃₁=y₃^Ty₁, and b=r₂₁y₂+r₃₁y₃;

computing γ

as

$\mathrm{\gamma <br><br>}=\min <br><br>\left(\frac{r_{21}^2+r_{31}^2}{b^T<br><br>b},\sqrt{\frac{\mathrm{\beta <br><br>}-1}{r_{21}^2+r_{31}^2}}\right);<br><br>$

and iii) computing I(r,c) as I(r,c)=∥

y₁−

γ

b∥

².

2. The method of claim 1 wherein the image data is N₁×

N₂ complex valued image data, the method further comprising the step of;

forming image domain looks by;

i) applying a FFT to convert the complex-valued image data into M₁×

M₂ frequency-domain data;

ii) removing any weighting function;

iii) generating a plurality of frequency domain looks as a plurality of M_L,1×

M_L,2 subsets of the M₁×

M₂ frequency domain data, where M_L,i is an integer near nM_i with 0<

n<

1, i=1,2; and

iv) transforming each frequency domain look into a corresponding image domain look using a 2M_L,1×

2M_L,2 inverse FFT, packing the data such that it is centered about zero frequency.

3. The method of claim 2, further comprising the step of:

for each image domain look, generating one beamspace look using a real part of the image domain look and generating a second beamspace look using an imaginary part of the image domain look.

4. The method of claim 1, further comprising the step of:

combining the high-definition image with an unweighted image and a Taylor weighted image to form a more detailed image by taking a pixel-by-pixel minimum of the high-definition image, unweighted image, and Taylor weighted image.

5. The method as per claim 4, wherein a filter is applied to sharpen the more detailed image.

6. The method of claim 1, wherein the image data is 2-D complex-valued image data.

7. The method of claim 6, wherein the 2-D image data is SAR data.

8. The method of claim 1, wherein the image data is 1-D image data.

9. The method of claim 8, wherein the 1-D image data is high range resolution profile data.

10. A system for processing image data to produce a high-definition image, comprising:

a peprocessing routine to receive the image data and generate a plurality of image domain looks;

a make beamspace looks routine to generate k beamspace looks, x₁ . . . x_k, from the plurality of image domain looks;

a minimum variance method routine to iteratively compute the high-definition image from the beamspace looks, wherein the high-definition image I is expressed in range and cross-range as I(r,c)=minω

^HRω

, where ω

is a weighting vector and R is a covariance matrix of the image data, wherein a solution for I(r,c) is approximated by i) forming Y=[x₁ . . . x_K]^T/{square root}{square root over (K)}, y₁, y₂, and y₃ denoting the K×

1 columns of Y;

ii) computing r₂₁=y₂^Ty₁ and r₃₁=y₃^Ty₁, and b=r₂₁y₂+r₃₁y₃;

computing γ

as

$\mathrm{\gamma <br><br>}=\min <br><br>\left(\frac{r_{21}^2+r_{31}^2}{b^T<br><br>b},\sqrt{\frac{\mathrm{\beta <br><br>}-1}{r_{21}^2+r_{31}^2}}\right);<br><br>$

and iii) computing I(r,c) as I(r,c)=∥

y₁−

γ

b∥

².

11. The system of claim 10 wherein the image data is N₁×

N₂ complex valued image data and the preprocessing routine forms image domain looks by;

i) applying a FFT to convert the complex-valued image data into M₁×

M₂ frequency-domain data;

ii) removing any weighting function;

iii) generating a plurality of frequency domain looks as a plurality of M_L,1×

M_L,2 subsets of the M₁×

M₂ frequency domain data, where M_L,i is an integer near nM_i with 0<

n<

1, i=1,2; and

iv) transforming each frequency domain look into a corresponding image domain look using a 2M_L,1×

2M_L,2 inverse FFT, packing the data such that it is centered about zero frequency.

12. The system of claim 11, wherein, for each image domain look, one beamspace look is generated using a real part of the image domain look and a second beamspace look is generated using an imaginary part of the image domain look.

13. The system of claim 10, further comprising:

an image combining routine to combine the high-definition image with an unweighted image and a Taylor weighted image to form a more detailed image by taking a pixel-by-pixel minimum of the high-definition image, unweighted image, and Taylor weighted image.

14. The system of claim 13, wherein the image combining routine applies a filter to sharpen the more detailed image.

15. The system of claim 10, wherein the image data is 2-D complex-valued image data.

16. The system of claim 15, wherein the 2-D image data is SAR data.

17. The system of claim 10, wherein the image data is 1-D data.

18. The system of claim 17, wherein the 1-D image data is high range resolution profile data.