System for topographic mapping from remotely sensed images

• US 5,926,581 A
• Filed: 04/25/1996
• Issued: 07/20/1999
• Est. Priority Date: 04/25/1996
• Status: Expired due to Fees

• Current Assignee: Lockheed Martin Corporation (Martin Marietta Corporation)
• Original Assignee: Lockheed Martin Corporation (Martin Marietta Corporation)
• Inventors: Pritt, Mark D.
• Primary Examiner(s): Johns, Andrew W.
• Assistant Examiner(s): Mariam, Daniel G.
• Application Number:
  US08/637,842
• Time in Patent Office: 1,181 Days
• Field of Search: 382/284, 382/294, 382/154, 348/145, 348/164, 348/159, 356/2
• US Class Current: 382/294
• CPC Class Codes: G01C 11/06 by comparison of two or more pictures of the same area
• CPC Class Codes: G06T 2200/08 involving all processing steps from image acquisition to 3D model generation
• CPC Class Codes: G06T 2207/10012 Stereo images
• CPC Class Codes: G06T 3/147 using affine transformations
• CPC Class Codes: G06T 7/593 from stereo images
• CPC Class Codes: H04N 13/239 using two 2D image sensors having a relative position equal to or related to the interocular distance H04N13/243 takes precedence
• CPC Class Codes: H04N 2013/0081 Depth or disparity estimation from stereoscopic image signals

Abstract

In a method for generating a model of terrain, images of the terrain are generated at different angles. Matching points in the images are identified and from these matching points, the coefficients of a registration mapping equation are determined. A height term of the registration matching equation is corrected for scaling and shear into an elevation term in the coordinates of one of the images. The elevation is rotated into the coordinates of the ground plane of the terrain to provide a digital elevation model of the terrain. Interpolation is carried out through the digital elevation model to provide a three-dimensional model of the terrain. In a system for measuring the accuracy of a registration mapping equation, first and second images are generated from a digital elevation model. The registration mapping equation being tested is applied to the images to generate an estimated digital matching function which is compared with the actual registration matching function for the two images.

Claims

1. A method of generating a digital elevation model of terrain comprising generating a first image of said terrain, generating a second image of said terrain at a different viewing angle than said first image, determining matching points in said first and second images, defining a registration matching equation which matches points from said first image to said second image, said matching equation including coefficients, determining from said matching points with a data processor the coefficients of said registration matching equation, said matching equation having a height term which varies with the terrain elevation in the coordinates of one of said images, correcting with said data processor said height term to correspond to the elevation of said terrain in the coordinates of one of said images, and rotating with said data processor the elevation of said terrain into the coordinates of a ground plane corresponding to said terrain to provide a digital elevation model.

2. A method of producing a three-dimensional model of terrain comprising performing the steps of the method of claim 1 and interpolating with said data processor through the digital elevation model to generate a three-dimensional model of said terrain.

3. A method as recited in claim 1, wherein said registration matching function comprises:

space="preserve" listing-type="equation">(u,v)=A(x,y)+he,

wherein x and y are the coordinates in the plane of one of said images, u and v are the coordinates in the plane of the other one of said images, A(x,y) is an affine transformation, h is a height term and e is an epipolar vector.

4. A method as recited in claim 3, wherein the coefficients of said registration matching equation are determined from said matching points by least squares.

5. A method as recited in claim 4, further comprising flagging the matching points which produce the highest fitting errors in the least squares determination and recomputing the registration matching function from a set of matched points which do not include the flagged matching points.

7. A method as recited in claim 3, wherein said height term is related to the elevation in one of said images in accordance with a predetermined function and wherein said height term is corrected to correspond to the elevation in the coordinates of one of said images in accordance with said predetermined function.

8. A method as recited in claim 7, wherein said predetermined function is:

space="preserve" listing-type="equation">z.sub.i =(h.sub.i -α

x.sub.i -By.sub.i)/γ

in which z_i is the elevation of the pixel i in the coordinates of said one of said images, x_i is the x coordinate of the pixel i in the coordinates of said one of said images, y_i is the y coordinate of the pixel i in the coordinates of said one of said images, and α

, B, and γ

are scaler values which depend on the relative orientation of the two images.

9. A method as recited in claim 8, wherein

10. A method as recited in claim 1 wherein said height term is related to the elevation in one of said images in accordance with a predetermined function and wherein said height term is corrected to correspond to the elevation in accordance with said predetermined function.

6. A system for measuring the accuracy of a registration mapping equation comprising means for providing a digital elevation model of terrain, means responsive to said digital elevation model to generate images of said digital elevation model at first and second viewing angles, means to apply a registration mapping equation to said first and second images to generate an estimated registration mapping function, means to produce the actual registration mapping function for said images, and means to generate a quantitative measurement of the difference between said estimated registration mapping function and said actual registration mapping function.