Granted Patent B2

US 12,563,310 · App. 18/597,798 · Granted Feb 24, 2026

Inventors: Kartik Venkataraman (San Jose, CA); Amandeep S. Jabbi (San Francisco, CA); Robert H. Mullis (Santa Cruz, CA); Jacques Duparre (Jena, DE); Shane Ching-Feng Hu (Fremont, CA)

Assignee: Adeia Imaging LLC

H04N23/90G02B3/0056G02B3/0062G02B5/20G02B5/201G02B13/0015G06T7/50G06T7/557G06T11/60G06T19/20H04N5/262H04N5/265H04N13/128H04N13/239H04N23/11H04N23/13H04N23/16H04N23/45H04N23/54H04N23/55H04N23/60H04N23/69H04N23/698H04N23/88H04N23/951H04N25/41H04N25/48H04N25/581H04N25/6153H04N25/705H04N25/79G06T2200/04G06T2207/10028G06T2207/10052G06T2207/20221H04N2013/0081H04N13/257H04N25/131H04N25/133H04N25/135

Abstract

Claims (25)

1 . A method of generating a depth map from a plurality of images captured from different viewpoints, comprising:

capturing a plurality of images from different viewpoints, where each image includes pixels that are occluded in at least one other image from the plurality of images;

normalizing the plurality of images based upon calibration data;

measuring parallax information between the normalized images by comparing similarity of neighborhoods of pixels for different parallax-induced shifts; and

generating a depth map using the measured parallax information.

2 . The method of claim 1 , wherein the plurality of images are captured by at least one camera array.

3 . The method of claim 2 , wherein the at least one camera array is a camera array of a mobile device.

4 . The method of claim 1 , wherein the plurality of images are captured by a plurality of cameras.

5 . The method of claim 4 , wherein the plurality of images include different occlusion sets, an occlusion set of a specific image captured by a given camera is a portion of a scene visible to a baseline camera from the plurality of cameras that is occluded from a viewpoint of the given camera; and wherein the plurality of cameras includes a first camera that captures pixels around an edge of a foreground object that is visible to the baseline camera and is in an occlusion set of a second of the plurality of cameras.

6 . The method of claim 4 , wherein cameras in the plurality of cameras are configured to operate with at least one difference in operating parameters.

7 . The method of claim 1 , wherein the plurality of images comprises a plurality of low resolution images, wherein the method further comprises generating at least one higher resolution super-resolved image using the plurality of the low resolution images by performing parallax detection to generate parallax information to align portions of the plurality of low resolution images.

8 . The method of claim 1 , wherein measuring parallax information comprises determining the parallax that yields a highest correlation between pixels from the plurality of images.

9 . The method of claim 1 , wherein measuring parallax information comprises calculating a parallax difference that yields a highest pixel correlation, wherein the parallax difference that yields the highest pixel correlation is determined by keeping track of various pair-wise measurements.

10 . The method of claim 1 , wherein measuring parallax information comprises performing pair-wise measurements of neighborhoods of pixels to determine pixel similarity for different parallax-induced shifts.

11 . The method of claim 10 , wherein measuring parallax information further comprises: determining a parallax that yields a highest similarity between pixels from images captured by keeping track of various pair-wise measurements of neighborhoods of pixels, and calculating a parallax that yields the highest similarity as a best least squares fit of the pair-wise measurements of neighborhoods of pixels.

12 . The method of claim 1 , further comprising generating at least one image by fusing aligned portions of the plurality of captured images using the depth map.

13 . The method of claim 12 , further comprising performing super-resolution processing on the fused image portions to synthesize a super-resolution image.

14 . The method of claim 1 , further comprising generating a higher resolution super-resolved image synthesized using the plurality of images and the parallax information to compensate for parallax in the plurality of images.

15 . The method of claim 1 , further comprising identifying occluded pixels based upon the measured parallax information.

16 . The method of claim 1 , wherein comparing the similarity of neighborhoods of pixels for different parallax-induced shifts comprises comparing the similarity of neighborhoods of pixels for different parallax-induced shifts along scan lines.

17 . The method of claim 1 , wherein normalizing the plurality of images based on calibration data comprises:

detecting and metering parallax using a parallax confirmation and measurement module, where detecting and metering parallax comprises ignoring pixels in images of the plurality of images that are in an exposed occlusion set;

aligning portions of images in the plurality of images captured by different cameras to compensate for parallax using an image pixel correlation module based upon the detected and metered parallax and stored calibration data; and

obtaining a higher resolution image having a resolution that is higher than resolutions of the images in the plurality of images using a super-resolution module, where color information around at least one edge of a foreground object that is visible in a baseline image and in an occlusion set of a second image is reconstructed in the higher resolution image using the pixels captured by a first image.

18 . The method of claim 1 , further wherein measuring parallax information comprises comparing a difference between average values of neighboring pixels with a threshold and flagging a potential presence of parallax when the difference exceeds the threshold.

Assignments (9)

Continuity (16)

References Cited (400)