Back to results
US 10,708,571 B2Granted
Video frame processing
AssigneeMicrosoft Technology Licensing, LLC
InventorsTomi Sokeila, Klaus Melakari, Mikko Muukki
Life of this patent
Estimating the term…
Abstract
An apparatus for or processing video frames is presented. The apparatus comprises at least one processing unit and at least one memory. The at least one memory stores program instructions that, when executed by the at least one processing unit, cause the apparatus to segment objects appearing in video frames, model movement of a camera recording the video frames, and individually compensate motion artefacts in the video frames for at least one segmented object based on relative movement of the camera and at least one segmented object.
Claims20
An apparatus, comprising:
- at least one processing unit;
- at least one memory;
- wherein the at least one memory stores program instructions that, when executed by the at least one processing unit, cause the apparatus to:
- segment objects appearing in video frames;
- model movement of a camera recording the video frames by:
- detecting a change in a position of selected points in a scene between successive video frames;
- and
- modeling a three-dimensional position of the camera in space using three-dimensional pose estimation based at least on the detected change in the position of the selected points;
- identifying motion artefacts for at least one segmented object in the video frames caused by a movement of the camera;
- individually correct one or more errors from the motion artefacts in the video frames for the at least one segmented object based on the modeled movement of the camera and the at least one segmented object;
- identifying one or more holes in the at least one segmented object based on individually correcting the one or more errors;
- and
- filling the one or more holes based on the data from previous or upcoming video frames.
10 more elements
A method comprising:
- segmenting objects appearing in video frames;
- modelling movement of a camera recording the video frames by:
- detecting a change in a position of selected points in a scene between successive video frames;
- and
- modeling a three-dimensional position of the camera in space using three-dimensional pose estimation based at least on the detected change in the position of the selected points;
- identifying motion artefacts for at least one segmented object in the video frames caused by a movement of the camera;
- individually correcting one or more errors from the motion artefacts in the video frames for that least one segmented object based on the modeled movement of the camera and at least one segmented object;
- identifying one or more holes in the at least one segmented object based on individually correcting the one or more errors;
- and
- filling the one or more holes based on the data from previous or upcoming video frames.
7 more elements
An apparatus, comprising:
- at least one processing unit;
- at least one memory;
- wherein the at least one memory stores program instructions that, when executed by the at least one processing unit, cause the apparatus to:
- obtain a file comprising video frames;
- obtain information relating to a camera that was used to record the video frames;
- segment objects appearing in the video frames;
- model movement of the camera based on at least the information relating to the camera by:
- detecting a change in a position of selected points in a scene between successive video frames;
- and
- modeling a three-dimensional position of the camera in space using three-dimensional pose estimation based at least on the detected change in the position of the selected points;
- identify motion artefacts for at least one segmented object in the video frames caused by a movement of the camera;
- individually correct one or more errors from the motion artefacts in the video frames for the at least one segmented object based on the modeled movement of the camera and at least one segmented object;
- identifying one or more holes in the at least one segmented object based on individually correcting the one or more errors;
- and
- filling the one or more holes based on the data from previous or upcoming video frames.
12 more elements
Dependencies are inferred from the claim text.