Semi-automatic extraction of linear features from image data

• US 7,653,218 B1
• Filed: 06/18/2007
• Issued: 01/26/2010
• Est. Priority Date: 05/02/2006
• Status: Active Grant

• Current Assignee: Maxar Intelligence Incorporated (Galileo Parent, Inc.)
• Original Assignee: GeoEye Solutions, Inc.
• Inventors: Grodecki, Jacek Franciszek, Bergen, Zachary, Poulsen, Richard, Lutes, James Anthony, Malitz, Seth M.
• Primary Examiner(s): Tarcza; Thomas H
• Assistant Examiner(s): Brainard; Timothy A
• Application Number:
  US11/764,765
• Time in Patent Office: 953 Days
• Field of Search: 342/191, 382/113, 382/154, 382/174, 382/257
• US Class Current: 382/113
• CPC Class Codes: G06V 20/13 Satellite images

Abstract

A method for extracting a linear feature from remotely sensed imagery, including selecting by user interface anchor points for the linear feature, the anchor points being identified with a geographic location; using image-based logic to automatically calculate a vector set associated with the anchor points, the vector set comprising a path associated with the linear feature; automatically attributing a material type to the path; automatically attributing a geometry to the path; selecting by user interface anchor points for a successive linear feature; using image-based logic to calculate a successive vector set associated with the anchor points for the successive linear feature, the successive vector set comprising a successive path; using a geometric relationship between the vector set and the successive vector set to automatically adjust the vector set to create an adjusted vector set, the adjusted vector set correcting the path associated with the first linear feature.

Claims

1. A method for extracting a linear feature from a remotely-sensed image, comprising:

providing access to a remotely-sensed image file containing image data from a multispectral image, a hyperspectral image, a radar image and a panchromatic image;

selecting by user interface the remotely-sensed image from the remotely-sensed image file;

selecting by user interface a plurality of first anchor points for the linear feature, the first anchor points being identified with a first geographic location;

using image-based logic to automatically calculate a first vector set associated with the first anchor points, the first vector set comprising a first path associated with the linear feature;

automatically attributing a material type to the first path;

automatically calculating a width for the first path;

selecting by user interface a second anchor point for a second linear feature, the second anchor point being identified with a second geographic location in the remotely-sensed image;

using image-based logic to automatically calculate a second vector set associated with the second anchor point, the second vector set comprising a second path; and

in real time, automatically using a geometric relationship between the first vector set and the second vector set to correct the first vector set or the second vector set, creating a corrected vector set, the corrected vector set resulting in a path correction to the first path or the second path.

2. The method of claim 1, wherein the automatically attributing a material type to the first path comprises using a material attribution algorithm and the material type comprises one selected from the group consisting of concrete, medium asphalt, dark asphalt, light unpaved, gravel and soil.

3. The method of claim 1, further comprising:

displaying the remotely-sensed image in a visual display; and

wherein using image-based logic to calculate the first vector set comprises connecting the first anchor points by a negative rubber band line in the visual display.

4. The method of claim 1, wherein, in real time, automatically using a geometric relationship comprises using an automatic vector revision function.

5. The method of claim 1, further comprising using the first vector set or the second vector set, the corrected vector set, and the width of the path correction to make a map.

6. The method of claim 1, further comprising preprocessing the remotely-sensed image by removing water pixels.

7. The method of claim 3, further comprising using the path correction and the material type to automatically make a map.

8. The method of claim 4, wherein the automatic vector revision function comprises at least one selected from the group consisting of automatic topology cleaning, automatic corner point installation and automatic smoothing.

9. The method of claim 4, wherein the automatic vector revision functions comprise at least one selected from the group consisting of automatic topology cleaning, automatic corner point installation and automatic smoothing.

10. A method for correcting an error in an extracted linear feature in a remotely-sensed image, comprising:

providing access to a remotely-sensed image file containing image data from a multispectral image, a hyperspectral image, a radar image and a panchromatic image;

selecting by user interface the remotely-sensed image from the remotely-sensed image file;

selecting by user interface a plurality of anchor points for a first linear feature in the remotely-sensed image, the plurality of anchor points being identified with a geographic location;

extracting the first linear feature by using image-based logic to automatically calculate a first vector set associated with the plurality of anchor points, the first vector set comprising a first path associated with the first linear feature;

selecting by user interface at least one additional anchor point for a second linear feature in the remotely-sensed image, the at least one additional anchor point being associated with a second geographic location;

extracting the second linear feature by using image-based logic to automatically calculate a second vector set associated with the at least one additional anchor point, the second vector set comprising a second path associated with the second linear feature;

selecting by user interface an attachment radius, the attachment radius being used to automatically define a region of influence around a center anchor point, the center anchor point being connected to the attachment radius and selected from among the plurality of anchor points for the first linear feature and the additional anchor point for the second linear feature;

using a geometric relationship between the first vector set and the second vector set within the region of influence to automatically identify the error, the error being associated with the first vector set or the second vector set; and

automatically revising the first vector set or the second vector set to correct the error and regenerate the first path or the second path, resulting in a regenerated first path or regenerated second path.

11. The method of claim 10, wherein automatically revising the first vector set or the second vector set comprises installing a corner, establishing an orthogonal crossroad, or removing a plurality of wiggles.

12. The method of claim 10, further comprising using the regenerated first path or regenerated second path to automatically generate a map.

13. The method of claim 10, wherein the region of influence comprises a line and node snapping region.

14. The method of claim 10, wherein the region of influence is expressed graphically and wherein the selecting by user interface the attachment radius comprises using a motion-sensitive device.

15. The method of claim 10, further comprising preprocessing the remotely-sensed image to remove water pixels.

16. A method for editing a vector set associated with a linear feature in a remotely-sensed image, comprising:

using image-based logic, automatically generating a path associated with the linear feature, the path comprising the vector set, the vector set being associated with at least a first user-selected anchor point and a second user-selected anchor point for the linear feature, the first user-selected anchor point and the second user-selected anchor point each being tied to a geographic location;

establishing a region of influence, the region of influence having the first user-selected anchor point or the second user-selected anchor point at its center;

selecting a point within the region of influence, the point having a vector, the region of influence being an area within which a geometric relationship between the vector and the vector set may be automatically evaluated; and

using the geometric relationship between the vector and the vector set to automatically make an adjustment to the vector set, the adjustment resulting in an adjusted path.

17. The method of claim 16, further comprising selecting an attachment radius, the attachment radius emanating from the first user-selected anchor point or the second user-selected anchor point around which the region of influence is centered.

18. The method of claim 16 further comprising:

automatically calculating a geometry for the path; and

automatically calculating an adjusted geometry for the adjusted path.

19. The method of claim 18 further comprising using the adjusted path and the vector to automatically make a map.

20. The method of claim 16 further comprising preprocessing the remotely-sensed image by generating a histogram or smoothing.

21. A method of editing a vector set, the vector set comprising a path associated with a linear feature in a graphic image, comprising:

by user interface, reviewing the path for an error;

identifying the error, the error being associated with the vector set;

selecting an editing tool appropriate to correct the error;

selecting by user interface an anchor point, the anchor point being in the vicinity of and logically related to the error and comprising an anchor point vector, the anchor point vector having a geometric relationship to the vector set;

establishing a region of influence, the region of influence encompassing the error and the anchor point, the region of influence being an area in which the geometric relationship can be used; and

automatically invoking the editing tool, thereby using the geometric relationship to adjust the vector set in real time to correct the error and create a revised path.

22. The method of claim 21, wherein the graphic image is a remotely-sensed image and the vector set and the anchor point vector set are each associated with a geographic location.

23. The method of claim 22, further comprising using the revised path to automatically make a map.

24. The method of claim 21, wherein reviewing the path for the error comprises using the region of influence to review the path for the error, and wherein the region of influence is expressed graphically.

25. The method of claim 21 wherein the editing tool comprises a corner break installation tool, a 1-point detour tool, an N-point detour tool, a move terminals tool, a smooth tool, a straighten tool, or a fuse tool.

26. A method for establishing a region of influence for a graphic image comprising a vector set, comprising:

using a motion-sensitive device to select a point in the graphic image, the motion-sensitive device being operatively associated with a visual display of the graphic image;

using the point and the motion-sensitive device to establish an attachment radius, the attachment radius being connected to the point;

using the attachment radius, automatically establishing the region of influence, the region of influence defining a subset of the graphic image, the subset including the vector set, being of a first size, centered about the point, and expressed graphically;

using the visual display, evaluating the subset of the graphic image, including the vector set, within the region of influence; and

automatically revising the vector set.

27. The method of claim 26, further comprising

using the motion-sensitive device to automatically adjust the attachment radius, changing the first size of the region of influence to a second size.

28. The method of claim 26, further comprising

using the motion-sensitive device to move the region of influence to center it about a second point in the graphic image.

29. A method for revising at least one vector set, the at least one vector set being one of a plurality of vector sets, the plurality of vector sets being tied to geographic locations and comprising a plurality of paths associated with a plurality of linear features in a remotely-sensed image, comprising:

preprocessing the remotely-sensed image by;

computing atmospheric correction and computing a multispectral cost file if the remotely-sensed image is a multispectral image;

computing a hyperspectral cost file if the remotely-sensed image is a hyperspectral image;

generating a texture file and computing a panchromatic cost file if the remotely-sensed image is a panchromatic image; and

smoothing the remotely-sensed image and computing a radar cost file if the remotely-sensed image is a radar image;

automatically evaluating a geometric relationship between the plurality of vector sets;

based on the evaluation, automatically determining whether an error exists in any of the plurality of vector sets;

automatically identifying the error;

automatically selecting an automatic vector revision tool to correct the error;

automatically correcting the error using the automatic vector revision tool, the correcting resulting in a revised vector set associated with a path; and

automatically using the revised vector set associated with the path to redraw the path.

30. The method of claim 29 further comprising using the revised vector set to automatically generate a map.

31. The method of claim 29, wherein the automatic vector revision tool comprises an automatic point snapping tool, an automatic orthogonal crossroads tool, an automatic corner installation tool or an automatic deep smoothing tool.

32. A machine readable medium having stored thereon instructions that, when executed by the machine, cause the machine to revise at least one vector set, the at least one vector set being one of a plurality of vector sets, the plurality of vector sets being tied to geographic locations and comprising a plurality of paths associated with a plurality of linear features in a remotely-sensed image, by:

preprocessing the remotely-sensed image according to image type;

automatically evaluating a geometric relationship between the plurality of vector sets;

based on the evaluation, automatically determining whether an error exists in any of the plurality of vector sets;

automatically identifying the error;

automatically selecting an automatic vector revision tool to correct the error;

automatically correcting the error using the automatic vector revision tool, the correcting resulting in a revised vector set; and

automatically using the revised vector set to redraw the path containing the revised vector set in real time.

33. The machine readable medium of claim 32 further comprising using the revised vector set to automatically generate a map.

34. The machine readable medium of claim 32 wherein the automatic vector revision tool comprises an automatic point snapping tool, an automatic orthogonal crossroads tool, or an automatic deep smoothing tool.

35. A method for extracting a linear feature from a remotely-sensed image, comprising:

providing access to a remotely-sensed image file containing a multispectral image, a panchromatic image, a hyperspectral image, and a radar image;

selecting the remotely-sensed image from the remotely-sensed image file;

preprocessing the remotely-sensed image and producing a preprocessed remotely-sensed image by;

computing atmospheric correction and computing a multispectral cost file if the remotely-sensed image is the multispectral image;

computing a hyperspectral cost file if the remotely-sensed image is the hyperspectral image;

generating a texture file and computing a panchromatic cost file if the remotely-sensed image is the panchromatic image; and

smoothing the remotely-sensed image and computing a radar cost file if the remotely-sensed image is the radar image;

displaying the preprocessed remotely-sensed image;

selecting by user interface a plurality of first anchor points for the linear feature, the plurality of first anchor points being identified with a first geographic location;

using image-based logic to automatically calculate a first vector set associated with the plurality of first anchor points, the first vector set comprising a first path associated with the linear feature;

selecting by user interface a second anchor point for a second linear feature, the second anchor point being identified with a second geographic location in the remotely-sensed image;

using image-based logic to automatically calculate a second vector set associated with the second anchor point, the second vector set comprising a second path;

establishing a region of influence around a center point, the center point being either the second anchor point or one of the plurality of first anchor points and connected to an attachment radius, the attachment radius and the center point defining the region of influence; and

automatically implementing vector revision functions to automatically;

evaluate the first vector set and the second vector set to determine if an error exists;

if the error exists, change the first vector set or the second vector set to correct the error, resulting in a changed vector set; and

revise the first path or the second path based on the changed vector set to produce a revised path.

36. The method of claim 35, further comprising displaying the revised path on the remotely-sensed image without displaying the first path and the second path.