Samsung patents video decoding that uses pixel gradients to improve motion prediction
Samsung has filed a patent for a video decoding method that calculates how pixels move by analyzing the brightness changes, or gradients, in reference images. The technique uses these gradients to predict pixel positions more accurately without needing extra data from outside the standard reference areas. This process helps reconstruct video frames by combining block-level and pixel-group-level motion compensation.
In the filing’s own words
Provided is a video decoding method including determining a displacement vector per unit time of pixels of a current block in a horizontal direction or a vertical direction, the pixels including a pixel adjacent to an inside of a boundary of the current block, by using values about reference pixels included in a first reference block and a second reference block, without using a stored value about a pixel located outside boundaries of the first reference block and the second reference block; and obtaining a prediction block of the current block by performing block-unit motion compensation and pixel group unit motion compensation on the current block by using a gradient value in the horizontal direction or the vertical direction of a first corresponding reference pixel in the first reference block which corresponds to a current pixel included in a current pixel group in the current block, a gradient value in the horizontal direction or the vertical direction of a second corresponding reference pixel in the second reference block which corresponds to the current pixel, a pixel value of the first corresponding reference pixel, a pixel value of the second corresponding reference pixel, and a displacement vector per unit time of the current pixel in the horizontal direction or the vertical direction. In this regard, the current pixel group may include at least one pixel.
What exactly they claim (claim 1)
1. A video decoding method by a video decoding apparatus comprising a processor, the video decoding method comprising: obtaining, from a bitstream, information related to a first motion vector and information related to a second motion vector; obtaining a first extended reference block in a first reference picture using the first motion vector, and a second extended reference block in a second reference picture using the second motion vector, wherein the first extended reference block comprises a first reference block and a first portion extended from the first reference block, and the second extended reference block comprises a second reference block and a second portion extended from the second reference block; determining a displacement vector of a pixel group of a current block, the pixel group comprising at least one pixel adjacent to an inside of a boundary of the current block, by using a gradient value of at least one reference pixel in the first extended reference block and a gradient value of at least one reference pixel in the second extended reference block, wherein the first portion of the first extended reference block is used to calculate a gradient value of at least one reference pixel in the first reference block, and the second portion of the second extended reference block is used to calculate a gradient value of at least one reference pixel in the second reference block; obtaining a predicted pixel value of the current block by performing optical flow based compensation on the current block by using the gradient value of the at least one reference pixel in the first reference block, the gradient value of the at least one reference pixel in the second reference block, and the displacement vector of the pixel group; and reconstructing the current block based on the predicted pixel value, wherein the pixel group comprises a K×K block, K being an integer greater than 1, wherein when a first intermediate value used for determining the displacement vector is greater than a predetermined value, the displacement vector is determined by performing a clipping operation on a second intermediate value based on a first threshold value and a second threshold value, the second intermediate value being derived using the first intermediate value.
Facts
- Company
- Samsung (SAMSUNG ELECTRONICS CO., LTD.)
- Official title
- VIDEO DECODING METHOD AND APPARATUS AND VIDEO ENCODING METHOD AND APPARATUS
- Publication number
- US20260292234A1
- Application number
- 19673593
- Filed
- May 11, 2026
- Published
- Sep 24, 2026
- Inventors
- 2
- Claims
- 3
Classified under
H04N 19/513 Electricity & electronicsH04N 19/105 Electricity & electronicsH04N 19/139 Electricity & electronicsH04N 19/159 Electricity & electronicsH04N 19/176 Electricity & electronicsMore from Samsung in AI & software
- Samsung patents tool to erase objects from photos using AI Sep 24, 2026
- Samsung patents drawing objects into photos with AI Oct 15, 2026
- Samsung patents AI that creates personalized exercise videos Sep 10, 2026
- Samsung patents AI summaries for paused TV shows based on your viewing habits Sep 24, 2026
- Samsung patents turning chat conversations into image prompts Oct 1, 2026
Summaries are simplified from the filing’s own abstract and first claim. A patent filing is not a product announcement.
Patent data and drawings: U.S. Patent and Trademark Office (USPTO). Nebula is not affiliated with or endorsed by the USPTO.