Method and system for simultaneously registering multi-dimensional topographical points

• US 7,567,731 B2
• Filed: 07/15/2004
• Issued: 07/28/2009
• Est. Priority Date: 07/15/2004
• Status: Active Grant

• Current Assignee: Harris Corporation (L3Harris Technologies, Inc.)
• Original Assignee: Harris Corporation (L3Harris Technologies, Inc.)
• Inventors: Fermo, Ray, Auxier, Jake, Rahmes, Mark, McDowall, Tom
• Primary Examiner(s): Mehta; Bhavesh M
• Assistant Examiner(s): Akhavannik; Hadi
• Application Number:
  US10/892,047
• Publication Number:
  US 20060013443A1
• Time in Patent Office: 1,839 Days
• Field of Search: 382/154, 382/294
• US Class Current: 382/294
• CPC Class Codes: G01S 3/7864 T.V. type tracking systems
• CPC Class Codes: G06T 7/35 using statistical methods

Abstract

A method for registering multi-dimensional topographical data points representing a multi-dimensional object, comprises: a) receiving a plurality of points representing a plurality of overlapping frames of a surface of the multi-dimensional object; b) finding for each point in a first frame a corresponding closest point in each of a plurality of subsequent frames; c) determining a rotation and translation transformation for each of the plurality of frames so that corresponding closest points are aligned; d) determining a cost for performing each rotation and translation transformation; and e) repeating steps b) through d) for additional frames to provide an optimum transformation.

Claims

1. A method for registering a set of frames of 3-dimensional digital elevation model data points which include both ground points representing a 3-dimensional surface and obstruction points which represent an obstruction lying above the 3-dimensional surface and are positionally interspersed among the ground points, said method comprising the steps of:

(a) extracting a ground surface shell from the set based on a plurality of ground points from which the obstruction points have been isolated;

(b) finding a plurality of pairs of corresponding closest points in overlapping frames of said ground surface shell such that a plurality of said pairs of said corresponding closest points are found for each one of a plurality of pairs of said overlapping frames;

(c) determining a plurality of 3-dimensional transformations based on said corresponding closest points found in step (a), each respective one of said transformations defining a 3-dimensional rotation and translation representing an alignment of a first frame and a second frame of a respective one of said overlapping pairs of frames, said transformations being determinable without reliance on ego motion information, and

(d) substantially simultaneously registering all frames included in said plurality of pairs of said overlapping frames by performing a rigid transform according to a matrix in which all of said transformations are represented.

2. The method of claim 1 further comprising the step of optimizing said pair-wise transformations.

3. The method of claim 2 wherein said optimizing step comprises the step of iteratively repeating steps (b) and (c).

4. The method of claim 1 wherein a Mahalanobis distance is used for determining which of said points in said overlapping frames constitute a said pair of corresponding closest points.

5. The method of claim 2 further comprising the step of determining a cost associated with research said pair-wise transformations and continuing said step of iteratively repeating said step of iteratively repeating steps (b) and (c) subject to an abort criterion related to said cost.

6. The method of claim 1 further comprising the step of carrying out an iterative process which comprises the step of iteratively repeating steps (b), (c) and (d).

9. The method of claim 1, wherein step (b) comprises the step of searching a k-D tree for finding said corresponding closest points.

10. The method of claim 1, wherein each of said points in said overlapping frames is represented by x, y, z coordinates and a Mahalanobis distance in the z direction is used to determine which of said points comprise said closest corresponding points.

11. The method of claim 1, further comprising the step of integrating said plurality of said frames after said registration step has been carried out to produce an integrated 3D model representing the 3-dimensional surface.

12. The method of claim 1, further comprising the step of displaying said integrated 3D model of the 3-dimensional surface.

13. The method of claim 11, wherein the ground points comprise both (i) 3-dimensional digital elevation model data points derived from a target partially obstructed by said obstruction and (ii) 3-dimensional digital elevation model topographical data points representing a geographical area in the vicinity of said target, and wherein said 3D model comprises a model which represents both said geographical area and said target.

14. The method of claim 3, further comprising the step of displaying said 3D model show the structure of said target.

7. A system for registering a set of frames of 3-dimensional digital elevation model points which include both ground points representing a 3-dimensional surface and obstruction points which represent an obstruction lying above the 3-dimensional surface and are positionally interspersed among the ground points, said system comprising:

a ground finder for isolating the obstruction points from the ground points;

a competitive filter for obtaining a ground surface shell from a plurality of ground points provided by said ground finder, said ground surface shell including a representation of each respective one of said frames;

a registration engine coupled to said competitive filter for (i) finding a plurality of pairs of corresponding closest points in overlapping frames of said ground surface shell such that a plurality of said pairs of said corresponding closest points are found for each one of a plurality of pairs of said overlapping frames, and (ii) determining a plurality of 3-dimensional transformations based on said corresponding closest points;

each respective one of said transformations defining a 3-dimensional rotation and translation representing an alignment of a first frame and a second frame of a respective one of said overlapping pairs of frames, said transformations being determinable without reliance on ego motion information, and

a transform engine coupled to said registration engine for substantially simultaneously registering all of said frames included in said plurality of pairs of said overlapping frames by performing a rigid transform according to a matrix in which all of said transformations are represented.

15. The system of claim 7, wherein said registration engine finds each said pair of corresponding closest points by searching a k-D tree.

16. The system of claim 7, further comprising a frame integrator for integrating said frames which have been registered by said transform engine to provide an integrated 3D model representing said 3-dimensional surface.

17. The system of claim 7, further comprising a display for displaying said 3D model representing said 3-dimensional surface.

8. A computer readable medium comprising program instructions for:

(a) extracting a ground surface shell from the set based on a plurality of ground points from which the obstruction points have been isolated;

(b) finding a plurality of pairs of corresponding closest points in overlapping frames of said ground surface shell such that a plurality of said pairs of said corresponding closest points are found for each one of a plurality of pairs of said overlapping frames;

(c) determining a plurality of 3-dimensional transformations based on said corresponding closest points found in step (a), each respective one of said transformations defining a 3-dimensional rotation and translation representing an alignment of a first frame and a second frame of a respective one of said overlapping pairs of frames, said transformations being determinable without reliance on ego motion information, and

(d) substantially simultaneously registering all frames included in said plurality of pairs of said overlapping frames by performing a rigid transform according to a matrix in which all of said transformations are represented.

18. The medium of claim 8, wherein instruction (b) comprises an instruction for organizing the points in each frame into a k-D tree structure and searching the k-D tree for each said corresponding closest 3-dimensional DEM point.

19. The medium of claim 8, a wherein each of said points in said overlapping frames is represented by x, y, z coordinates and a Mahalanobis distance in the z direction is used to determine which of said points comprise said closest corresponding points.