Method of solving initial azimuth for survey instruments, cameras, and other devices with position and tilt information

• US 9,891,049 B2
• Filed: 10/29/2015
• Issued: 02/13/2018
• Est. Priority Date: 10/29/2015
• Status: Active Grant

• Current Assignee: Trimble Inc.
• Original Assignee: Trimble Inc.
• Inventors: Brown, Troy Lane
• Primary Examiner(s): Perungavoor, Sath V
• Assistant Examiner(s): Brown, Jr., Howard D
• Application Number:
  US14/926,394
• Publication Number:
  US 20170122733A1
• Time in Patent Office: 838 Days
• CPC Class Codes: G01C 11/02 Picture taking arrangements specially adapted for photogrammetry or photographic surveying, e.g. controlling overlapping of pictures
• CPC Class Codes: G01C 11/04 Interpretation of pictures
• CPC Class Codes: G06T 2207/30181 Earth observation
• CPC Class Codes: G06T 2207/30244 Camera pose
• CPC Class Codes: G06T 7/33 using feature-based methods
• CPC Class Codes: G06T 7/55 from multiple images
• CPC Class Codes: G06T 7/73 using feature-based methods
• CPC Class Codes: G06T 7/97 Determining parameters from multiple pictures depth or shape recovery from multiple images G06T7/55; stereo camera calibration G06T7/85
• CPC Class Codes: H04N 23/698 for achieving an enlarged field of view, e.g. panoramic image capture
• CPC Class Codes: H04N 23/90 Arrangement of cameras or camera modules, e.g. multiple cameras in TV studios or sports stadiums

Abstract

A method for solving the initial azimuth for a survey instrument, and other devices, with position and tilt information. The method may be part of bundle adjustment performed during the processing of image and position data collected from each camera of a survey instrument. The use of the initial azimuth generation method (and initial azimuth generator) makes it possible to get accurate azimuth orientation of a camera, such as each camera of a multi-camera survey instrument, without using data from a compass-like sensor. The initial azimuth generated by this method can then be used in later steps/processes of the bundle adjustment to find tie-points with an automatic tie-point finder. Prior to this method, the automatic tie-point finding algorithm relied on the compass and its accuracy, and inaccurate compass values would cause a complete failure, slow runtimes, or less accurate results for the bundle adjustment.

Claims

1. A method for processing data from a camera-based device with improved initial azimuth accuracy, comprising:

providing a dataset comprising digital images captured by a camera, position information for the camera when each of the digital images was captured, and tilt information for the camera when each of the digital images was captured;

generating a plurality of potential positions for tie-points for the dataset;

performing arbitration to select a most-likely tie-point position from the plurality of potential tie-point positions associated with an initial tie-point;

generating a plurality of additional tie-points based on the most-likely tie-point position of the initial tie-point and an initial azimuth of the camera provided by the initial tie-point; and

performing bundle adjustment on the dataset using the initial tie-point and the plurality of additional tie-points.

2. The method of claim 1, wherein subsets of the digital images are captured at two or more differing locations in a physical environment.

3. The method of claim 2, wherein the digital images are captured at four differing locations in the physical environment, wherein the method further comprises receiving a manually or automatically generated tie-point and wherein the generating of the plurality of potential positions for the tie-points includes, with the manually or automatically generated tie-point as input, applying a four-station-one-tie-point algorithm to the dataset to generate a first subset of the plurality of potential positions for the tie-points.

4. The method of claim 3, further comprising:

identifying four cameras that observe the manually or automatically generated tie-point;

with an initial pose for each of the four cameras, computing lens position and a direction vector for an observation of the tie-point in a world frame; and

creating a hyperbolic cone based on the lens position and the direction vector for each of the four cameras.

5. The method of claim 4, where the first subset of the plurality of potential positions for the tie-points is determined by finding intersections of the hyperbolic cones.

6. The method of claim 3, wherein the digital images are captured at two differing locations in the physical environment, wherein the method further comprises receiving two manually or automatically generated tie-points, and wherein the generating of the plurality of potential positions for the tie-points includes, with the two manually or automatically generated tie-points as input, applying a two-station-two-tie-point algorithm to the dataset to generate a second subset of the plurality of potential positions for the tie-points.

7. The method of claim 6, further comprising:

identifying two cameras at two different ones of the locations of the physical environment that were used to capture the digital images that both observe the two manually or automatically generated tie-points;

with an initial pose for each of the two cameras, computing lens position and a direction vector for an observation of the tie-point in a world frame; and

creating a hyperbolic cone based on the lens position and the direction vector for each of the two cameras.

8. The method of claim 4, where the second subset of the plurality of potential positions for the tie-points is determined by finding intersections of the hyperbolic cones associated with the two cameras.

9. The method of claim 1, wherein the most-likely tie-point position is a tie-point position that produces an azimuth that most closely matches an azimuth provided by data from a compass sensor associated with the camera.

10. A photogrammetry method, comprising:

positioning a device with two or more cameras, a position sensor, and a tilt sensor at a plurality of differing locations;

operating the device to capture at each of the locations a dataset including digital images, position data, and tilt data;

applying a four-station-one-tie-point algorithm to the dataset to generate a first list of potential tie-point positions each associated with a tie-point for the dataset;

applying a two-station-one-tie-point algorithm to the dataset to generate a second list of potential tie-point positions each associated with a tie-point for the dataset; and

processing the first and second lists of potential tie-point positions to select a tie-point at a most-likely reliable tie-point position.

11. The method of claim 10, wherein the first list of potential tie-point positions correspond with intersections of hyperbolic cones generated from portions of the dataset captured at four of the differing locations.

12. The method of claim 11, wherein the four of the differing locations are selected as all providing observations of a tie-point.

13. The method of claim 10, wherein the second list of potential tie-point positions correspond with intersections of hyperbolic cones generated from portions of the dataset captured at two of the differing locations.

14. The method of claim 13, wherein the two of the differing locations are selected as both providing observations of a pair of tie-points.

15. The method of claim 10, further comprising computing initial azimuth values for each of the cameras at each of the differing locations based on the selected tie-point and initiating a bundle adjustment on the dataset based on the selected tie-point and the computed initial azimuth values.

16. A system for processing geodata from camera-based devices to determine initial azimuth, comprising:

a camera-based device comprising at least one camera operable to capture digital images of a physical environment from two or more locations in the physical environment, wherein the camera-based device further includes position and tilt sensors capturing position and tilt information for the camera-based device concurrently with operation of the camera to capture the digital images; and

a server providing a geodata analysis module to first process the captured digital images and the position and tilt information in combination with a manually or automatically generated tie-point to generate a plurality of potential tie-point positions, to second determine a most-likely tie point from the plurality of potential tie-point positions, and to provide a bundle adjustment of data gathered by the camera-based device using the most-likely tie point and an initial azimuth for the camera associated with the most-likely tie point.

17. The system of claim 16, wherein the geodata analysis module comprises a four-station-one-tie-point algorithm generating hyperbolic cones using the manually or automatically generated tie-point and four cameras from which the manually or automatically generated tie-point is observable from different ones of the locations and wherein the plurality of potential tie-point positions correspond with intersections of the hyperbolic cones.

18. The system of claim 16, wherein the geodata analysis module comprises a two-station-two-tie-point algorithm generating hyperbolic cones using the manually or automatically generated tie-point, an additional manually or automatically generated tie-point, and two cameras from which both of the generated tie-points are observable from two different ones of the locations and wherein the plurality of potential tie-point position correspond with intersections of the hyperbolic cones.

19. The system of claim 16, wherein the camera-based device comprises a survey instrument with a plurality of cameras that are concurrently operable to capture with the digital images a 360-degree panoramic view of the physical environment.

20. The system of claim 19, wherein an initial azimuth is determined for each of the plurality of cameras by the geodata analysis module for use as input for the bundle adjustment.