Target-less auto-alignment of image sensors in a multi-camera system

• US 9,792,667 B2
• Filed: 11/07/2016
• Issued: 10/17/2017
• Est. Priority Date: 03/20/2014
• Status: Expired due to Fees

• Current Assignee: GoPro, Inc.
• Original Assignee: GoPro, Inc.
• Inventors: Sun, Yajie, Macmillan, Timothy, Edwards, Elizabeth Torres, Chawla, Sumit
• Primary Examiner(s): Natnael, Paulos M
• Application Number:
  US15/345,002
• Publication Number:
  US 20170053392A1
• Time in Patent Office: 344 Days
• Field of Search: 348 47, 348 46, 348 48, 348 26, 348625, 382266, 382294, 382103, 382199, 382296
• CPC Class Codes: G06F 18/22 Matching criteria, e.g. proximity measures
• CPC Class Codes: G06T 3/4038 Image mosaicing, e.g. composing plane images from plane sub-images
• CPC Class Codes: G06T 5/70 Denoising; Smoothing
• CPC Class Codes: G06T 7/13 Edge detection
• CPC Class Codes: G06T 7/337 involving reference images or patches
• CPC Class Codes: G06T 7/80 Analysis of captured images to determine intrinsic or extrinsic camera parameters, i.e. camera calibration
• CPC Class Codes: G06T 7/85 Stereo camera calibration
• CPC Class Codes: H04N 13/239 using two 2D image sensors having a relative position equal to or related to the interocular distance H04N13/243 takes precedence
• CPC Class Codes: H04N 13/243 using three or more 2D image sensors
• CPC Class Codes: H04N 13/246 Calibration of cameras
• CPC Class Codes: H04N 13/271 wherein the generated image signals comprise depth maps or disparity maps
• CPC Class Codes: H04N 13/296 Synchronisation thereof; Control thereof
• CPC Class Codes: H04N 17/002 for television cameras
• CPC Class Codes: H04N 2013/0081 Depth or disparity estimation from stereoscopic image signals
• CPC Class Codes: H04N 23/689 Motion occurring during a rolling shutter mode
• CPC Class Codes: H04N 23/80 Camera processing pipelines; Components thereof
• CPC Class Codes: H04N 25/48 Increasing resolution by shifting the sensor relative to the scene
• CPC Class Codes: H04N 25/76 Addressed sensors, e.g. MOS or CMOS sensors

Abstract

A pair of cameras having an overlapping field of view is aligned based on images captured by image sensors of the pair of cameras. A pixel shift is identified between the images. Based on the identified pixel shift, a calibration is applied to one or both of the pair of cameras. To determine the pixel shift, the camera applies correlation methods including edge matching. Calibrating the pair of cameras may include adjusting a read window on an image sensor. The pixel shift can also be used to determine a time lag, which can be used to synchronize subsequent image captures.

Claims

1. A computer-implemented method for determining a pixel shift between an image pair captured by image sensors, the method comprising:

accessing a first image and a second image of the image pair captured at a substantially same time, the images comprising image data representative of an overlapping field of view between the image sensors;

performing a smoothing operation on the first image and the second image to produce smoothed image data;

determining edge magnitude of the smooth image data based on derivatives along at least a row and a column of pixel luma components in the smoothed image data;

determining edge phase of the smoothed image data based on an inverse tangent of the derivatives of the pixel luma components;

identifying, by one or more processors, one or more edges in the image data based, at least in part, on the edge magnitude and the edge phase of the smoothed image data;

matching the identified one or more edges in the image data corresponding to the first image to the identified one or more edges in the image data corresponding to the second image; and

determining a pixel shift between the image pair based, at least in part, on the matching of edges.

2. The method of claim 1, wherein identifying the one or more edges in the image data comprises:

identifying a first set of smoothed image data having determined edge magnitude greater than an edge magnitude threshold;

identifying a second set of smoothed image data having determined edge phase within a threshold of a target edge phase value; and

identifying the one or more edges from image data in both the first set of smoothed image data and the second set of smoothed image data.

3. The method of claim 2, wherein the identified one or more edges are substantially perpendicular to a rolling shutter direction of the image sensors, and the target edge phase value corresponds to the rolling shutter direction.

4. The method of claim 1, wherein matching the identified one or more edges in the image data corresponding to the first image to the identified one or more edges in the image data corresponding to the second image comprises:

determining edge lengths for the identified one or more edges; and

matching the one or more edges based, at least in part, on the determined edge magnitude of the one or more edges and the determined edge lengths.

5. The method of claim 1, wherein the image sensors have a rolling shutter direction, the method further comprising:

before determining the edge magnitude and determining the edge phase, compressing the first and second images perpendicular to the rolling shutter direction without compressing the first and second images along the rolling shutter direction.

6. The method of claim 1, wherein the first image and the second image are downsampled before the smoothing operation is performed.

7. The method of claim 1, further comprising cropping the first and second images by removing image data corresponding to outer borders of the first and second images prior to determining the edge magnitude and determining the edge phase.

8. The method of claim 1, further comprising calibrating the image sensors using the determined pixel shift.

9. A non-transitory computer readable medium comprising instructions for determining a pixel shift between an image pair captured by image sensors, the instructions executable by a processor, the instructions comprising instructions for:

accessing a first image and a second image of the image pair captured at a substantially same time, the images comprising image data representative of an overlapping field of view between the image sensors;

performing a smoothing operation on the first image and the second image to produce smoothed image data;

determining edge magnitude of the image data based on derivatives along at least a row and a column of pixel luma components in the smoothed image data;

determining edge phase of the smoothed image data based on an inverse tangent of the derivatives of the pixel luma components;

identifying, by one or more processors, one or more edges in the image data based, at least in part, on the edge magnitude and the edge phase of the smoothed image data;

matching the identified one or more edges in the image data corresponding to the first image to the identified one or more edges in the image data corresponding to the second image; and

determining a pixel shift between the image pair based, at least in part, on the matching of edges.

10. The computer-readable medium of claim 9, wherein identifying the one or more edges in the image data comprises:

identifying a first set of smoothed image data having determined edge magnitude greater than an edge magnitude threshold;

identifying a second set of smoothed image data having determined edge phase within a threshold of a target edge phase value; and

identifying the one or more edges from image data in both the first set of smoothed image data and the second set of smoothed image data.

11. The computer-readable medium of claim 10, wherein the identified one or more edges are substantially perpendicular to a rolling shutter direction of the image sensors, and the target edge phase value corresponds to the rolling shutter direction.

12. The computer-readable medium of claim 9, wherein matching the identified one or more edges in the image data corresponding to the first image to the identified one or more edges in the image data corresponding to the second image comprises:

determining edge lengths for the identified one or more edges; and

matching the one or more edges based, at least in part, on the determined edge magnitude of the one or more edges and the determined edge lengths.

13. The computer-readable medium of claim 9, wherein the image sensors have a rolling shutter direction, wherein the instructions further comprise instructions for:

before determining the edge magnitude and determining the edge phase, compressing the first and second images perpendicular to the rolling shutter direction without compressing the first and second images along the rolling shutter direction.

14. The computer-readable medium of claim 9, wherein the instructions further comprise instructions for calibrating the image sensors using the determined pixel shift.

15. A camera system comprising at least one processor, an image sensor, and a non-transitory computer readable medium comprising instructions for determining a pixel shift between an image pair captured by image sensors including the image sensor, the instructions executable by the at least one processor, the instructions comprising instructions for:

accessing a first image and a second image of the image pair captured at a substantially same time, the images comprising image data representative of an overlapping field of view between the image sensors;

performing a smoothing operation on the first image and the second image to produce smoothed image data;

determining edge magnitude of the image data based on derivatives along at least a row and a column of pixel luma components in the smoothed image data;

determining edge phase of the smoothed image data based on an inverse tangent of the derivatives of the pixel luma components;

identifying, by one or more processors, one or more edges in the image data based, at least in part, on the edge magnitude and the edge phase of the smoothed image data;

matching the identified one or more edges in the image data corresponding to the first image to the identified one or more edges in the image data corresponding to the second image; and

determining a pixel shift between the image pair based, at least in part, on the matching of edges.

16. The camera system of claim 15, wherein identifying the one or more edges in the image data comprises:

identifying a first set of smoothed image data having determined edge magnitude greater than an edge magnitude threshold;

identifying a second set of smoothed image data having determined edge phase within a threshold of a target edge phase value; and

identifying the one or more edges from image data in both the first set of smoothed image data and the second set of smoothed image data.

17. The camera system of claim 16, wherein the identified one or more edges are substantially perpendicular to a rolling shutter direction of the image sensors, and the target edge phase value corresponds to the rolling shutter direction.

18. The camera system of claim 15, wherein matching the identified one or more edges in the image data corresponding to the first image to the identified one or more edges in the image data corresponding to the second image comprises:

determining edge lengths for the identified one or more edges; and

matching the one or more edges based, at least in part, on the determined edge magnitude of the one or more edges and the determined edge lengths.

19. The camera system of claim 15, wherein the image sensors have a rolling shutter direction, wherein the instructions further comprise instructions for:

before determining the edge magnitude and determining the edge phase, compressing the first and second images perpendicular to the rolling shutter direction without compressing the first and second images along the rolling shutter direction.

20. The camera system of claim 15, wherein the instructions further comprise instructions for calibrating the image sensors using the determined pixel shift.