Method and apparatus maintaining eye contact in video delivery systems using view morphing
AssigneeApplied Minds, LLC
InventorsW. Daniel Hillis, Bran Ferren, Russel Howe
Life of this patent
Estimating the term…
Abstract
A view morphing algorithm is applied to synchronous collections of video images from at least two video imaging devices, and interpolating between the images, creates a composite image view of the local participant. This composite image approximates what might be seen from a point between the video imaging devices, presenting the image to other video session participants.
Claims50
An apparatus for presenting a composite image conveying eye contact with a conference participant, comprising:
- a composite image generator communicatively coupled with at least two members of an imaging device collection, for repeatedly receiving an image from each image device in said collection at approximately a same time to provide a sychromized image collection of said images based upon observations of said participant by said at least two members of said imaging device collection;
- wherein said composite image generator is communicatively coupled to a motion video portal for providing a succession of said composite images based upon at least said synchronized image collection for presentation to a video delivery system;
- wherein said video delivery system presents a second participant with a motion video stream conveying eye contact generated by said motion video portal based upon said composite image succession;
- wherein said composite image generator is composite of;
- means for calculating at least one dense correspondence to determine a displacement in at least a first dimension for each pixel in a first image for one of said imaging device collection members which would move each of said pixels to a most nearly corresponding pixel in a second image for at least one other of said imaging device collection members;
- and
- means for generating an interpolated image from each of at least two of said synchronized image collection members, wherein said interpolated image is comprised of an array of pixels.
Method for conveying eye contact with a participant in a video conference session comprising the steps of:
- obtaining at least two images from each of at least two imaging devices;
- wherein each of said images is comprised of a two-dimensional array of pixels of approximately the same size and shape;
- calculating a dense correspondence to determine a displacement in a first dimension for each of said pixels in at least one first image which would move each of said pixels therein to a most nearly corresponding pixel in said image of at least one other of said imaging devices;
- and generating an interpolated image for each of at least two of said imaging devices wherein each of said interpolated images is comprised of essentially said array of pixels.
A method of conveying eye contact with a local participant to at least one second participant, comprising the steps of:
- calculating at least one dense correspondence to determine a displacement in at least a first dimension for each pixel in a first image produced by a first imaging device which would move each of said pixels to a most nearly corresponding pixel in an image produced by another imaging device that is displaced from said first imaging device relative to said local participant;
- generating an interpolated image comprising essentially an array of pixels for each of said imaging devices.
A method of conveying eye contact with a local participant to at least one second participant, comprising the steps of:
- calculating at least one dense correspondence to determine a displacement in at least a first dimension for each pixel in a first image produced by a first imaging device which would move each of said pixels to a most nearly corresponding pixel in at least a second image produced by at least a second imaging device that is displaced from said first imaging device relative to said local participant;
- generating an interpolated image comprising essentially an array of pixels from each of said imaging devices;
- and
- adaptively modifying subsequent images, at least in part based upon previous displacements.
Dependencies are inferred from the claim text.