Image processing system for estimating camera parameters

• US 8,264,546 B2
• Filed: 09/11/2009
• Issued: 09/11/2012
• Est. Priority Date: 11/28/2008
• Status: Expired due to Fees

• Current Assignee: Sony Corporation (Sony Group Corp.)
• Original Assignee: Sony Corporation (Sony Group Corp.)
• Inventors: Witt, Sarah Elizabeth
• Primary Examiner(s): HERNANDEZ, NELSON D
• Application Number:
  US12/557,925
• Publication Number:
  US 20100134634A1
• Time in Patent Office: 1,096 Days
• Field of Search: 348/181, 348/180, 348/187, 348/188, 348/222.1, 382/285, 382/286, 382/288, 382/287, 382/289
• US Class Current: 348/187
• CPC Class Codes: G06T 7/74 involving reference images or patches
• CPC Class Codes: H04N 5/2224 related to virtual studio applications

Abstract

An image processing system includes a camera, an image processor and a calibration surface including a calibration pattern, comprising plural alternately colored elements which provide a corresponding plurality of corners at locations at which more than two of the colored elements adjoin. The image processor detects from the video signal at least some of the corner locations and identifies a first and second group of lines of the calibration pattern, extrapolate the lines from each group and determine a on a plane of the scene as viewed by the camera, the presence and location of a first and second intersection points on where the extrapolated lines from the first and second groups, respectively, intersect to estimate one or more of a roll pitch and yaw angle of the camera relative to the calibration surface and extimates a likelihood that a corner is located at each of identified potential corner locations.

Claims

1. An image processing system comprising:

a camera;

an image processor; and

a calibration surface including thereon a calibration pattern, the calibration pattern including a plurality of alternately colored elements which provide a corresponding plurality of corners at locations on the calibration pattern at which more than two of the colored elements adjoin, the corner locations defining a first group of lines and a second group of lines, the lines in each of the group of lines being parallel with respect to each other on calibration surface, the first group of lines being orthogonal to the second group of lines, wherein

the camera is configured to generate a video signal representative of a scene as viewed by the camera, the scene including the calibration surface,

the image processor is configured

to detect from the video signal at least some of the corner locations to identify the first and second group of lines of the calibration pattern,

to extrapolate the lines from each group of lines to determine a presence and a location of a first intersection point on a plane of the scene as viewed by the camera at which the extrapolated lines from the first group intersect and a second intersection point on the plane of scene as viewed by the camera at which the extrapolated lines from the second group intersect, and

to estimate from the presence and the location of the first and second intersection points one or more of an estimated roll angle value, an estimated pitch angle value and an estimated yaw angle value corresponding to a roll angle, pitch angle and yaw angle of the camera relative to the calibration surface,

the detection of the corner locations by the image processor includes identifying from the video signal a plurality of potential corner locations, defining an area of the scene as viewed by the camera centered around each potential corner location, analyzing a color of points lying on a periphery of each area, to estimate a likelihood that a corner is located at each potential corner location, and

the analyzing of the color of the points lying on the periphery of the area includes dividing the periphery into a first part and a second part, and comparing a distribution of color of the points on the first part with a distribution of color of the points on the second part.

2. The image processing system according to claim 1, wherein each point on the video signal is identified as a potential corner location.

3. The image processing system according to claim 1, wherein the analyzing of the color of the points lying on the periphery of the area includes identifying an estimated corner location at the potential corner location if the distribution of color of the points on the first part and the distribution of color of the points on the second part are substantially similar.

4. The image processing system according to claim 3, wherein the detection of the corner locations includes refining the estimated corner location by

defining a further area around the potential corner location,

determining edge points around the periphery of the further area at which two of the alternately colored elements meet, and

identifying a refined corner location at a point in the centre center of the area at which lines plotted between the edge points intersect.

5. The image processing system according to claim 1, wherein the corner locations are a predetermined distance apart on the calibration surface, the image processor being configured to determine an estimated center point of the calibration surface from a distribution of the corner locations over the calibration pattern and to determine an estimated X displacement value, an estimated Y displacement value and an estimated Z displacement, the estimated X, Y and Z displacement values corresponding to a position in three dimensional space of a position of the camera relative to the calibration surface.

6. The image processing system according to claim 5, wherein the image processor is configured to detect from the video signal representative of the scene at least one of a plurality of markers, each marker being uniquely identifiable by the image processor and occupying a predetermined position on the calibration surface, the image processor being configured to refine at least one of the estimate of the center point and the estimate of the yaw angle based on the at least one detected marker.

7. The image processing system according to claim 1, wherein the calibration pattern is a checkerboard pattern including alternately colored squares.

8. The image processing system according to claim 1, wherein if the extrapolated lines from one of the first or second groups of lines do not all intersect at a point, the image processing apparatus is configured to determine that one of either the roll angle is zero with respect to the calibration surface, or the pitch angle of the camera is normal with respect to the calibration surface.

9. The image processing system according to claim 1, wherein if the extrapolated lines from the first group of lines do not all intersect at a point and the extrapolated lines from the second group of lines do not all intersect at a point, the image processing apparatus is configured to determine that the roll angle is zero and the pitch angle of the camera is normal with respect to the calibration surface.

10. The image processing system according to claim 1, wherein the image processor is configured to generate a composite image including image data derived from the video signal combined with a rendering of a computer generated object, the computer generated object rendered in accordance with the estimated roll angle value, pitch angle value and yaw angle value.

11. The image processing system according to claim 1, wherein the video signal includes a plurality of video frames and the image processor is configured to estimate one or more of the estimated roll angle value, the estimated pitch angle value and the estimated yaw angle value for individual video frames of the video signal.

12. An image processor that receives a video signal from a camera representative of a scene as viewed by the camera, the scene including a calibration surface including thereon a calibration pattern, the calibration pattern including a plurality of alternately colored elements which provide a corresponding plurality of corners at locations on the calibration pattern at which more than two of the colored elements adjoin, the corner locations defining a first group of lines and a second group of lines, the lines in each of the group of lines being parallel with respect to each other and on the calibration surface, the first group of lines being orthogonal to the second group of lines, the image processor comprising:

a detection unit configured to detect from the video signal at least some of the corner locations to identify the first and second group of lines of the calibration pattern;

an extrapolation unit configured to extrapolate the lines from each group of lines to determine a presence and a location of a first intersection point on a plane of the scene as viewed by the camera at which the extrapolated lines from the first group intersect and a second intersection point on the plane of scene as viewed by the camera at which the extrapolated lines from the second group intersect; and

an estimating unit configured to estimate from the presence and the location of the first and second intersection points one or more of an estimated roll angle value, an estimated pitch angle value and an estimated yaw angle value corresponding to a roll angle, pitch angle and yaw angle of the camera relative to the calibration surface, wherein

the detection of the corner locations by the detection unit includes identifying from the video signal a plurality of potential corner locations, defining an area of the scene as viewed by the camera centered around each potential corner location, analyzing a color of points lying on a periphery of each area to estimate a likelihood that a corner is located at each potential corner location, and

the analyzing of the color of the points lying on the periphery of the area includes dividing the periphery into a first part and a second part, and comparing a distribution of color of the points on the first part with a distribution of color of the points on the second part.

13. A method of estimating camera parameters from a captured image, the method comprising:

generating a video signal representative of a scene as viewed by a camera, the scene including a calibration surface including thereon a calibration pattern, the calibration pattern including a plurality of alternately colored elements which provide a corresponding plurality of corners at locations on the calibration pattern at which more than two of the colored elements adjoin, the corner locations defining a first group of lines and a second group of lines, the lines in each of the group of lines being parallel with respect to each other and with respect to the calibration surface, the first group of lines being orthogonal to the second group of lines;

detecting from the video signal at least some of the corner locations to identify the first and second group of lines of the calibration pattern;

extrapolating the lines from each group of lines;

determining from the extrapolated lines a presence and a location of a first intersection point on a plane of the scene as viewed by the camera at which the extrapolated lines from the first group intersect and a second intersection point on the plane of the scene as viewed by the camera at which the extrapolated lines from the second group intersect; and

estimating from the presence and the location of the first and second intersection points one or more of an estimated roll angle value, an estimated pitch angle value and an estimated yaw angle value corresponding to a roll angle, pitch angle and yaw angle of the camera relative to the calibration surface, wherein

the detecting of the corner locations, includes

identifying from the video signal a plurality of potential corner locations,

defining an area of the scene as viewed by the camera centered around each potential corner location, and

analyzing a color of points lying on a periphery of each area to estimate a likelihood that a corner is located at each potential corner location, the analyzing of the color of the points lying on the periphery of the area including dividing the periphery into a first part and a second part, and comparing a distribution of color of the points on the first part with a distribution of color of the points on the second part.

14. A computer graphics generation system, comprising:

a camera configured to generate video images of a scene captured by the camera, the scene including a calibration surface including thereon a calibration pattern, the calibration pattern including a plurality of alternately colored elements which provide a corresponding plurality of corners at locations on the calibration pattern at which more than two of the colored elements adjoin, the corner locations defining a first group of lines and a second group of lines, the lines in each of the group of lines being parallel with respect to each other and on the calibration surface, the first group of lines being orthogonal to the second group of lines; and

an image processor configured to receive the video signal from the camera, and to combine the video images with rendered computer generated objects, the computer generated objects being rendered in accordance with a position and orientation of the camera determined by

detecting from the video signal at least some of the corner locations to identify the first and second group of lines of the calibration pattern,

extrapolating the lines from each group of lines to determine a presence and a location of a first intersection point on a plane of the scene as viewed by the camera at which the extrapolated lines from the first group intersect and a second intersection point on the plane of scene as viewed by the camera at which the extrapolated lines from the second group intersect, and

estimating from the presence and the location of the first and second intersection points one or more of an estimated roll angle value, an estimated pitch angle value and an estimated yaw angle value corresponding to a roll angle, pitch angle and yaw angle of the camera relative to the calibration surface, wherein

the detection of the corner locations by the image processor includes identifying from the video signal a plurality of potential corner locations, defining an area of the scene as viewed by the camera centered around each potential corner location, analyzing a color of points lying on a periphery of each area to estimate a likelihood that a corner is located at each potential corner location, and

the analyzing of the color of the points lying on the periphery of the area includes dividing the periphery into a first part and a second part, and comparing a distribution of color of the points on the first part with a distribution of color of the points on the second part.

15. A non-transitory computer readable storage medium having computer executable instructions, which when loaded on to a computer cause the computer to perform a method of estimating camera parameters from a captured image, the method comprising:

generating a video signal representative of a scene as viewed by a camera, the scene including a calibration surface including thereon a calibration pattern, the calibration pattern including a plurality of alternately colored elements which provide a corresponding plurality of corners at locations on the calibration pattern at which more than two of the colored elements adjoin, the corner locations defining a first group of lines and a second group of lines, the lines in each of the group of lines being parallel with respect to each other and with respect to the calibration surface, the first group of lines being orthogonal to the second group of lines;

detecting from the video signal at least some of the corner locations to identify the first and second group of lines of the calibration pattern;

extrapolating the lines from each group of lines;

determining from the extrapolated lines a presence and a location of a first intersection point on a plane of the scene as viewed by the camera at which the extrapolated lines from the first group intersect and a second intersection point on the plane of the scene as viewed by the camera at which the extrapolated lines from the second group intersect; and

estimating from the presence and the location of the first and second intersection points one or more of an estimated roll angle value, an estimated pitch angle value and an estimated yaw angle value corresponding to a roll angle, pitch angle and yaw angle of the camera relative to the calibration surface, wherein

the detecting of the corner locations includes

identifying from the video signal a plurality of potential corner locations,

defining an area of the scene as viewed by the camera centered around each potential corner location, and

analyzing a color of points lying on a periphery of each area to estimate a likelihood that a corner is located at each potential corner location, the analyzing of the color of the points lying on the periphery of the area including dividing the periphery into a first part and a second part, and comparing a distribution of color of the points on the first part with a distribution of color of the points on the second part.

16. An apparatus for estimating camera parameters from a captured image, the apparatus comprising:

means for generating a video signal representative of a scene as viewed by a camera, the scene including a calibration surface including thereon a calibration pattern, the calibration pattern including a plurality of alternately colored elements which provide a corresponding plurality of corners at locations on the calibration pattern at which more than two of the colored elements adjoin, the corner locations defining a first group of lines and a second group of lines, the lines in each of the group of lines being parallel with respect to each other and with respect to the calibration surface, the first group of lines being orthogonal to the second group of lines;

means for detecting from the video signal at least some of the corner locations to identify the first and second group of lines of the calibration pattern;

means for extrapolating the lines from each group of lines;

means for determining from the extrapolated lines a presence and a location of a first intersection point on a plane of the scene as viewed by the camera at which the extrapolated lines from the first group intersect and a second intersection point on the plane of the scene as viewed by the camera at which the extrapolated lines from the second group intersect; and

means for estimating from the presence and the location of the first and second intersection points one or more of an estimated roll angle value, an estimated pitch angle value and an estimated yaw angle value corresponding to a roll angle, pitch angle and yaw angle of the camera relative to the calibration surface, wherein

the means for detecting the corner locations includes

means for identifying from the video signal a plurality of potential corner locations,

means for defining an area of the scene as viewed by the camera centered around each potential corner location, and

means for analyzing a color of points lying on a periphery of each area to estimate a likelihood that a corner is located at each potential corner location, the means for analyzing the color of the points lying on the periphery of the area including means for dividing the periphery into a first part and a second part, and means for comparing a distribution of color of the points on the first part with a distribution of color of the points on the second part.