Granted Patent B2

US 12,591,957 · App. 18/343,589 · Granted Mar 31, 2026

Inventors: Debi Prasanna Mohanty (Karnataka, IN); Sukumar Moharana (Karnataka, IN); Siddhant Garg (Karnataka, IN); Kadaru Sai Ravi Teja (Karnataka, IN); Siva Prasad Thota (Karnataka, IN); Vanraj Vala (Karnataka, IN); Dwaraka Bhamidipati Sreevatsa (Karnataka, IN)

Assignee: SAMSUNG ELECTRONICS CO., LTD.

G06T5/80G06T3/60G06T5/50G06T7/70G06T2207/10016G06T2207/20081G06T2207/20132G06T2207/30168

Abstract

Claims (79)

1 . A method for tilt correction of a video by an electronic device, the method comprising:

receiving a user selection of the video;

identifying a sequence of image frames of the video representing a scene in the video;

determining a degree of tilt in each image frame of the sequence of image frames;

determining whether the degree of tilt in each image frame meets a first threshold condition using a trained Artificial Intelligence (AI) model;

determining that cause of tilt in the sequence of image frames is un-intentional based on the degree of tilt in each image frame meeting the first threshold condition; and

performing the tilt correction on the sequence of image frames based on the cause of tilt being un-intentional.

2 . The method as claimed in claim 1 , wherein performing the tilt correction on the sequence of image frames, comprises:

estimating a tilt angle of each image frame for maximum tilt correction at each image frame;

determining a tilt angle deviation of each image frame, wherein the tilt angle deviation of an image frame of the sequence of image frames is an absolute value of difference between the tilt angle of the image frame and a mean of the tilt angles of the image frame and an adjacent image frame of the sequence of image frames;

estimating a pixel shift and loss in each image frame and a region to crop on each image frame based on the tilt angle deviation and the tilt angle;

determining a target tilt correction required for each image frame based on the pixel shift and loss;

applying the target tilt correction, and at least one of cropping and out-painting on the determined region of each image frame;

stitching each tilt corrected image frame to rebuild the scene; and

generating a video at least by sequentially joining stitched image frames.

3 . The method as claimed in claim 2 , wherein estimating the tilt angle of each image frame for the maximum tilt correction at each image frame, comprises:

predicting a range of possible tilt angles of each image frame using a trained self-attention model; and

determining the tilt angle at least by determining a weighted mean of the predicted tilt angles with a standard deviation of zero or a larger value.

4 . The method as claimed in claim 3 , wherein a self-attention model is trained for tilt angle prediction at least by:

determining a range of possible tilt angles of training image frames based on relative position of objects in the training image frames, relevance of the objects, and relationship between the objects;

generating training data using the range of possible tilt angles;

generating an output vector with distributed probability at least by performing multi-label training on the training data; and

training the self-attention model with the output vector using a sigmoid activation for tilt angle prediction.

5 . The method as claimed in claim 2 , wherein estimating the pixel shift and loss in each image frame and the region to crop on each image frame based on the tilt angle deviation and the tilt angle, comprises:

identifying a minimum tilt angle of the sequence of image frames based on the tilt angle of each image frame;

determining a mean tilt angle deviation of the sequence of image frames based on the tilt angle deviation of each image frame;

determining whether the minimum tilt angle is greater than the mean tilt angle deviation;

performing at least one of:

setting the mean tilt angle deviation to the minimum tilt angle, and estimating an angle to correct tilt of each image frame, based on the minimum tilt angle being greater than the mean tilt angle deviation, and

estimating the angle to correct tilt of each image frame, based on the minimum tilt angle being not greater than the mean tilt angle deviation; and

determining the region to crop on each image frame based on the mean tilt angle deviation, wherein the target tilt correction is with minimum pixel shift and loss.

6 . The method as claimed in claim 5 , wherein estimating the angle to correct tilt of each image frame, comprises:

determining whether the tilt angle of an image frame of the sequence of image frames is one of equal to zero, smaller than or equal to mean tilt angle deviation, and greater than then mean tilt angle deviation; and

performing at least one of:

estimating the angle to correct tilt of the image frame as zero, based on the tilt angle of the image frame being equal to zero,

estimating the angle to correct tilt of the image frame as the tilt angle, based on the tilt angle of the image frame being smaller than or equal to the mean tilt angle deviation, and

estimating the angle to correct tilt of the image frame as the mean tilt angle deviation, based on the tilt angle of the image frame being greater than the mean tilt angle deviation.

7 . The method as claimed in claim 2 , wherein estimating the pixel shift and loss in each image frame and the region to crop on each image frame based on the tilt angle deviation and the tilt angle, comprises:

identifying a maximum tilt angle of the sequence of image frames based on the tilt angle of each image frame;

determining a mean tilt angle deviation of the sequence of image frames based on the tilt angle deviation of each image frame;

determining whether the tilt angle of an image frame of the sequence of image frames is one of equal to zero, and smaller than or equal to the maximum tilt angle;

performing at least one of:

estimating the angle to correct tilt of the image frame as zero, based on the tilt angle of the image frame being equal to zero, and

estimating the angle to correct tilt of the image frame as the tilt angle, based on the tilt angle of the image frame being smaller than or equal to the maximum tilt angle; and

determining the region to crop on each image frame based on the mean tilt angle deviation, wherein the target tilt correction is with maximum pixel shift and loss.

8 . The method as claimed in claim 2 , wherein determining the target tilt correction required for each image frame based on the pixel shift and loss, comprises:

determining whether a maximum tilt angle and a mean tilt angle deviation of the sequence of image frames meet a second threshold condition based on the pixel shift and loss; and

performing at least one of:

the target tilt correction with maximum pixel shift and loss, based on the maximum tilt angle and the mean tilt angle deviation of the sequence of image frames meeting the second threshold condition, and

the target tilt correction with minimum pixel shift and loss, based on the maximum tilt angle and the mean tilt angle deviation of the sequence of image frames do not meeting the second threshold condition.

9 . The method as claimed in claim 2 , wherein applying the target tilt correction and cropping to each image frame on the determined region of each image frame, comprises:

rotating each image frame based on an estimated angle to correct tilt of the image frame; and

cropping the determined region on each image frame based on the mean tilt angle deviation.

10 . The method as claimed in claim 2 , wherein applying at least one of cropping and out-painting on the determined region of each image frame, comprises:

determining whether a region to crop on each image frame meets a third threshold condition; and

performing at least one of:

cropping the sequence of image frames, based on the region to crop on each image not meeting the third threshold condition, and

cropping and out-painting the sequence of image frames, based on the region to crop on each image meeting the third threshold condition.

11 . An electronic device for tilt correction of a video, comprises:

a memory;

at least one processor comprising processor circuitry;

a tilt correction controller comprising processing circuitry, coupled to the memory and the at least one processor, the tilt correction controller configured for:

receiving a user selection of the video,

identifying a sequence of image frames of the video representing a scene in the video,

determining a degree of tilt in each image frame of the sequence of image frames,

determining whether the degree of tilt in each image frame meets a first threshold condition using a trained Artificial Intelligence (AI) model;

determining that cause of tilt in the sequence of image frames is un-intentional based on the degree of tilt in each image frame meeting the first threshold condition; and

performing the tilt correction on the sequence of image frames based on the cause of tilt being un-intentional.

12 . The electronic device as claimed in claim 11 , wherein performing the tilt correction on the sequence of image frames, comprises:

estimating a tilt angle of each image frame for maximum tilt correction at each image frame;

determining a tilt angle deviation of each image frame, wherein the tilt angle deviation of an image frame of the sequence of image frames is an absolute value of difference between the tilt angle of the image frame and a mean of the tilt angles of the image frame and an adjacent image frame of the sequence of image frames;

estimating a pixel shift and loss in each image frame and a region to crop on each image frame based on the tilt angle deviation and the tilt angle;

determining a target tilt correction required for each image frame based on the pixel shift and loss;

applying the target tilt correction, and at least one of cropping and out-painting on the determined region of each image frame;

stitching each tilt corrected image frame to rebuild the scene; and

generating a video at least by sequentially joining stitched image frames.

13 . The electronic device as claimed in claim 12 , wherein estimating the tilt angle of each image frame for the maximum tilt correction at each image frame, comprises:

predicting a range of possible tilt angles of each image frame using a trained self-attention model; and

determining the tilt angle at least by determining a weighted mean of the predicted tilt angles with a standard deviation of zero or a larger value.

Assignments (1)

Priority Claims (1)

Continuity (2)

References Cited (39)