Method of generating visual representation of terrain height from SAR data employing multigrid analysis

• US 5,608,405 A
• Filed: 10/06/1995
• Issued: 03/04/1997
• Est. Priority Date: 10/06/1995
• Status: Expired due to Term

• Current Assignee: Lockheed Martin Corporation (Martin Marietta Corporation)
• Original Assignee: Lockheed Martin Corporation (Martin Marietta Corporation)
• Inventors: Pritt, Mark D.
• Primary Examiner(s): Lobo, Ian J.
• Application Number:
  US08/539,928
• Time in Patent Office: 515 Days
• Field of Search: 342/25, 342/190, 342/191
• US Class Current: 342/25.00C
• CPC Class Codes: G01S 13/9023 combined with interferometric techniques

Abstract

In a method of recovering data representing terrain height from a synthetic aperture radar system the phase data is unwrapped by a Gauss-Seidel relaxation technique applied to a least squares differential equation in a multigrid algorithm. In the multigrid algorithm the problem represented by the least squares equation is transferred to successively coarser grids to more quickly remove the low frequency components of error. The resulting interim solutions determined on the coarser grids are then transferred successively back to the finer grids to converge on the finest grid of the grid array.

Claims

1. A method of generating a digital elevation model of a surface comprising transmitting radar pulses from a synthetic aperture radar system to be reflected from said surface,

receiving radar pulses reflected from said surface at a first angle and a second angle slightly different than said first angle with said synthetic aperture radar system,

performing the following steps with a data processor:

processing said received reflected radar pulses into a first set of data representing the phase of the received reflected pulses at said first angle from each increment of said surface and into a second set of data representing the phase of the reflected radar pulses at said second angle from each increment of said surface,

determining a third set of data representing the difference in the phases between said first set of data and said second set of data for each of said increments, said third set of data representing the relative position of increments of said surface in modulo form, unwrapping said third set of data into a fourth set of digital data representing the relative position of said surface increments in digital form, said unwrapping step comprising determining partial derivatives of said third set of data and representing said partial derivatives as a fifth set of data,

defining a least squares equation representing said fourth set of data in terms of said fifth set of data and the partial derivatives of said fourth set of data,

solving said equation by relaxation in a multigrid algorithm which converges to said fourth set of data.

2. A method as recited in claim 1, wherein said multigrid algorithm comprises defining a multiplicity of grids decreasing progressively from grids of finest resolution to grids of coarsest resolution, initially representing said fourth set of data on said grids of finest resolution, transferring the problem represented by said equation successively to coarser grids by means of a restriction operator, performing relaxation on said coarser grids to determine interim solutions on said coarser grids, and transferring the interim solutions determined on said coarser grids successively to finer grids by means of a prolongation operator to converge to said fourth set of data on said finest grid.

3. A method as recited in claim 1, further comprising generating an image of said surface represented by said fourth set of data.

4. A method as recited in claim 1, wherein the terms of said equation are weighted to take into account the reliability of said third set of data in representing the relative position of increments of said surface.

5. A method as recited in claim 4, wherein the terms of said equation are weighted in accordance with weighting values separately determined for each grid point of the finest grid of a pyramid of grids comprising successively coarser grids, said multigrid algorithm applying said weighting values to said successively coarser grids of said pyramid in accordance with a weight restriction operator, said weight restriction operator setting the weight value in the next coarser grid to be zero depending on the zero and nonzero status of neighboring grid positions of the preceding finer grid in said pyramid.

6. A method as recited in claim 1 wherein said third set of data is represented on a grid having an x and y axes and wherein said partial derivatives include d/dx partials and d/dy partials, wherein said d/dx partials in said equation are weighted by a first set of weighting coefficients derived from a first set of weighting values and wherein said d/dy partials in said equation are weighted by a second set of weighting coefficients derived from a second set of weighting values and wherein said multigrid algorithm restricts said first and second weighting values to coarser grids on separate grid arrays in accordance with a weight restriction operator, whereby corresponding weighting values can have different values on said separate grid arrays.

7. A method as recited in claim 6 wherein the weighting values on a given grid are represented by σ

_ij and τ

_ij and wherein σ

_ij represents the weighting values in the ith column and the jth row for the d/dx partials and wherein τ

_ij represents the weighting values in the ith column and the jth row for the d/dy partials and wherein the weighting value σ

_ij on a given coarse grid is set to be zero by said weight restriction operator if at least one of the following conditions hold;

(1) the two weighting values on the next adjacent finer grid σ

_2i,2j and σ

_2i,2j+1 are zero, (2) the weighting values on the next adjacent finer grid σ

_2i-1,2j a and σ

_2i,2j+1 are zero and the weighting value on such given coarse grid σ

_i-1,j is nonzero, or (3) the weighting value on the next adjacent finer grid σ

_2i,2j and σ

_2i-1,2j+1 are zero and the weighting value on such given coarse grid σ

_i-1,j is nonzero and wherein said weight restriction operator sets the weighting value τ

_ij on a given coarse grid to be zero if at least one of the following three conditions hold;

(1) the two weighting values on the next adjacent finer grid τ

_2i,2j and τ

_2i,2j+1 are zero, (2) the two weighting values on the next adjacent finer grid τ

_2i-1,2j and τ

_2i,2j+1 are zero and the weighting value on such given coarse grid τ

_i-1,j is nonzero, or (3) the weighting value on the next adjacent finer grid τ

_2i,2j and τ

_2i-1,2j+1 are zero and the weighting value on such given coarse grid τ

_i-1,j is nonzero.

8. A method as recited in claim 2 wherein said prolongation operator determines the values for a coarser grid by means of a prolongation operator value and adds the determined values to corresponding values on the next finer grid.

9. A method as recited in claim 8 wherein said prolongation operator is defined in accordance with the following equation:

##EQU5## wherein c_ij is a point on a grid in the ith column and the jth row and f_2i,2j is the corresponding point on the next finer grid in the 2ith column and 2jth row.

10. A method as recited in claim 2 wherein said restriction operator multiplies neighboring values by selected constants and adds the resulting products together to set a corresponding value on the next coarser grid.

11. A method as recited in claim 10 wherein each point on a coarse grid is set in accordance with the following equation:

##EQU6## wherein c_ij is a point on a grid in the ith column and the jth row and f_2i,2j is the corresponding point in the next finer grid in the 2ith column and the 2jth row.

12. A method as recited in claim 2 wherein after data representing partial derivatives are stored on a given grid, the first column of said given grid is set to equal the negative of the second column of said given grid and the first row of said given grid is set to equal the negative of the second row of said given grid.

13. A method as recited in claim 1 wherein said relaxation comprises Gauss-Seidel relaxation.