Determining the position of an object in a scene

Patent metadata

Field	Value
Publication number	US-10410068-B2
Application number	US-201715611816-A
Country	US
Kind code	B2
Filing date	Jun 2, 2017
Priority date	Jun 29, 2016
Publication date	Sep 10, 2019
Grant date	Sep 10, 2019

Abstract

A method of determining the position of an object in a scene, comprising: receiving captured images of the scene, each image being captured from a different field of view of the scene, wherein a portion of the scene with a volume comprises a detectable object, the volume is divided into volume portions, and each volume portion is within the captured field of view of at least two of the captured images so that an image of each volume portion appears in the at least two of the captured images; detecting, for each volume portion in each of the captured images within which an image of that volume portion appears, whether or not an image of one of the detectable objects in the scene is positioned within a distance of the position of the image of that volume portion, a correspondence between the images of the detectable objects detected in the at least two of the images is established, the correspondence indicating that the images of the detectable objects detected in the at least two of the images correspond to a single detectable object in the scene, and the position in the scene of that volume portion is established as a position in the scene of the single detectable object.

First claim

The invention claimed is:

1. A method of determining a position of an object in a scene, comprising:
receiving a plurality of captured images of the scene, each respective one of the plurality of images being captured from a different field of view of the scene, wherein a predetermined portion of the scene with a predetermined volume comprises a plurality of detectable objects, the predetermined volume is divided into a plurality of volume portions, and each volume portion is within the captured field of view of at least two of the captured images so that an image of each volume portion appears in the at least two of the captured images;
detecting, for each volume portion in each of the captured images within which an image of that volume portion appears, whether or not an image of one of the detectable objects in the scene is positioned within a predetermined distance of the position of the image of that volume portion, wherein if an image of one of the detectable objects is detected to be positioned within a predetermined distance of the position of the image of that volume portion in at least two of the images within which an image of that volume portion appears, then a correspondence between the images of the detectable objects detected in the at least two of the images is established, the correspondence indicating that the images of the detectable objects detected in the at least two of the images correspond to a single detectable object in the scene, and the position in the scene of that volume portion is established as a position in the scene of the single detectable object.

2. The method according to claim 1 , wherein the position in the scene of each volume portion that is established as a position in the scene of a single detectable object is established as an initial position of that single detectable object, wherein the initial position established for each single detectable object is used as an input, together with one or more initial parameters of each camera used for capturing the plurality of images, for performing bundle adjustment for finding a more accurate position of each single detectable object in the scene and a more accurate one or more initial parameters of each camera.

3. The method according to claim 1 , wherein the detectable objects in the scene comprise one or more light markers each projected from a light source.

4. The method according to claim 1 , wherein the detectable objects in the scene comprise one or more physical markers.

5. The method according to claim 1 , wherein the predetermined portion of the scene with the predetermined volume comprises a first object to be modelled and a plurality of the detectable objects are positioned on the first object, and wherein the method comprises generating a model of the first object on the basis of an established position of each detectable object in the scene.

6. The method according to claim 5 , wherein the first object is a chassis of a motor vehicle, and the method comprises determining at least one of a distance measurement and an angle measurement associated with the chassis of the motor vehicle on the basis of at least one of a corresponding distance measurement and a corresponding angle measurement, respectively, of the generated model of the chassis of the motor vehicle.

7. The method according to claim 5 , wherein the model of the first object is compared to a previously generated model of a second object.

8. The method according to claim 7 , wherein the first object is an outer body of a first motor vehicle and the second object is an outer body of a second motor vehicle, and the method comprises determining a distance between a point on the model of the outer body of the first motor vehicle and a corresponding point on the model of the outer body of the second vehicle.

9. The method according to claim 5 , wherein the first object is a sports playing area.

10. The method according to claim 9 , wherein:
the sports playing area comprises one or more additional objects which are characteristic of the sports playing area, each of the one or more additional objects being detectable in at least two of the captured images;
an initial position in the scene of each of the additional objects is provided on the basis of a known characteristic of the sports playing area; and
the initial position provided for each of the additional objects is used as an input, together with one or more parameters of each camera used for capturing the plurality of images, for performing bundle adjustment for finding a more accurate position of each of the additional objects in the scene.

11. An apparatus for determining a position of an object in a scene, comprising:
receiver circuitry operable to receive a plurality of captured images of the scene, each respective one of the plurality of images being captured from a different field of view of the scene, wherein a predetermined portion of the scene with a predetermined volume comprises a plurality of detectable objects, the predetermined volume is divided into a plurality of volume portions, and each volume portion is within the captured field of view of at least two of the captured images so that an image of each volume portion appears in the at least two of the captured images;
image processor circuitry operable to detect, for each volume portion in each of the captured images within which an image of that volume portion appears, whether or not an image of one of the detectable objects in the scene is positioned within a predetermined distance of the position of the image of that volume portion, wherein if an image of one of the detectable objects is detected to be positioned within a predetermined distance of the position of the image of that volume portion in at least two of the images within which an image of that volume portion appears, then a correspondence between the images of the detectable objects detected in the at least two of the images is established by the image processor circuitry, the correspondence indicating that the images of the detectable objects detected in the at least two of the images correspond to a single detectable object in the scene, and the position in the scene of that volume portion is established by the image processor circuitry as a position in the scene of the single detectable object.

12. The apparatus according to claim 11 , wherein the position in the scene of each volume portion that is established as a position in the scene of a single detectable object is established as an initial position of that single detectable object, wherein the initial position established for each single detectable object is used as an input, together with one or more initial parameters of each camera used for capturing the plurality of images, for the image processor circuitry to perform bundle adjustment for finding a more accurate position of each single detectable object in the scene and a more accurate one or more initial parameters of each camera.

13. The apparatus according to claim 11 , wherein the detectable objects in the scene comprise one or more light markers each projected from a light source.

14. The apparatus according to claim 11 , wherein the detectable objects in the scene comprise one or more physical markers.

15. The apparatus according to claim 11 , wherein the predetermined portion of the scene with the predetermined volume comprises a first object to be modelled and a plurality of the detectable objects are positioned on the first object, and wherein the image processor circuitry is operable to generate a model of the first object on the basis of an established position of each det

Assignees

• Sony Corp
• Sony Europe Ltd

Inventors

• Dodd Luke
• Hawkins Paul

Classifications

• G01C11/04

  Interpretation of pictures · CPC title

• G01S5/16 Primary

  using electromagnetic waves other than radio waves · CPC title

• G06T2207/10012

  Stereo images · CPC title

• G06T2207/30108

  Industrial image inspection · CPC title

• G06T7/593 Primary

  from stereo images · CPC title