# US 2026/0065443 A1 · Method and apparatus for video restoration

> Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt

- Number: US 2026/0065443 A1
- Document ID: US20260065443A1
- Status: Published application
- Kind: A1
- Application number: 19/028,322
- Filed: 2025-01-17
- Published: 2026-03-05
- Earliest claimed: 2024-09-04
- Assignees: Electronics and Telecommunications Research Institute
- Applicants: Electronics and Telecommunications Research Institute
- Inventors: Ji Wan KIM; Do Hyung Kim
- Web page: https://patentgrep.com/patent/US20260065443A1

## CPC

- G06T 5/77: Retouching; Inpainting; Scratch removal
- G06T 5/50: using two or more images, e.g. averaging or subtraction
- G06T 11/10: Texturing; Colouring; Generation of textures or colours (retouching, inpainting or scratch removal G06T5/77)
- G06T 2207/10016: Video; Image sequence
- G06T 2207/10024: Color image
- G06T: IMAGE DATA PROCESSING OR GENERATION, IN GENERAL

## Abstract

A video restoration method includes: receiving a target video to restore and a source video for restoration; registering the source video with the target video; converting both videos from RGB to YUV channels; identifying areas with significant differences in Y channel values as light reflection highlights; restoring these areas by replacing the Y channel values in the target video with corresponding values from the registered source video; and converting the restored target video back from YUV to RGB channels to produce the final restored video.

## Claims

### Claim 1

A video restoration method, comprising:
receiving a target video to be restored and a source video to be used for restoration of the target video;
registering the source video with reference to the target video;
changing the target video and the registered source video from an RGB channel to a YUV channel;
designating and selecting an area with a larger difference in Y channel value between the target video and the registered source video as an area in which light reflection highlight occurs;
restoring the area in which the light reflection highlight occurs by replacing a Y channel value of the area in which the light reflection highlight occurs in the target video with a Y channel value of a corresponding area of the registered source video; and
changing the target video in which the area in which the light reflection highlight occurs has been restored, from a YUV channel to an RGB channel to generate a restored video.

### Claim 2 (depends on 1)

The video restoration method of claim 1, wherein the target video and the source video are acquired using a multi-view video.

### Claim 3 (depends on 1)

The video restoration method of claim 1, wherein the restoring of the area in which the light reflection highlight occurs in the target video includes adjusting the luminance of the Y channel value of the corresponding area in the registered source video using an interpolation scheme and replacing the Y channel value of the area in the target video with the adjusted luminance value.

### Claim 4 (depends on 3)

The video restoration method of claim 3, wherein the interpolation scheme uses a formula:
Δ
⁢
p
~
(
i
)
=
1
e
-
s
⁢
(
(
e
-
i
)
⁢
Δ
⁢
p
⁡
(
s
)
+
(
i
-
s
)
⁢
Δ
⁢
p
⁡
(
e
)
)
(wherein, pixel domains p and q denote Y channel values of the source video and the target video, respectively, an i-th pixel value of the source video is p(i), an i-th estimated pixel value of the target video is {tilde over (q)}(i), {tilde over (q)}(i)=p(i)+Δ{tilde over (p)}(i) a start coordinate index of an area with a large difference in Y channel value is s, and an end coordinate index is e, Δp(s)=|q(s)−p(s)|, and Δp(e)=|q(e)−p(e)|).

### Claim 5 (depends on 4)

The video restoration method of claim 4, wherein a formula
Δ
⁢
p
˜
(
i
)
=
1
2
⁢
k
2
⁢
∑
j
⁢
(
Δ
(
j
)
+
Δ
(
j
)
)
(wherein, {tilde over (q)}(i) values of a row and a column are (i)=pr (i)+Δ(i) and (i)=pc (i)+Δ(i), respectively, and k indicates a window size including an area with a large difference in Y channel value) is used to apply the interpolation scheme to each of pixels of all rows and columns of the area in which the light reflection highlight occurs.

### Claim 6 (depends on 1)

The video restoration method of claim 1, further comprising:
restoring the area in which the light reflection highlight occurs in the source video by using the target video as a source video for the source video when there is the area in which the light reflection highlight occurs in the source video.

### Claim 7

A video restoration apparatus comprising at least one memory configured to store instructions, and at least one processor, the at least one processor executing the instructions to perform the processes of:
receiving a target video to be restored and a source video to be used for restoration of the target video;
registering the source video with reference to the target video;
changing the target video and the registered source video from an RGB channel to a YUV channel;
designating and selecting an area with a larger difference in Y channel value between the target video and the registered source video as an area in which light reflection highlight occurs;
restoring the area in which the light reflection highlight occurs by replacing a Y channel value of the area in which the light reflection highlight occurs in the target video with a Y channel value of a corresponding area of the registered source video; and
changing the target video in which the area in which the light reflection highlight occurs has been restored, from a YUV channel to an RGB channel to generate a restored video.

### Claim 8 (depends on 7)

The video restoration apparatus of claim 7, wherein the target video and the source video are acquired using a multi-view video.

### Claim 9 (depends on 7)

The video restoration apparatus of claim 7, wherein the process of restoring the area in which the light reflection highlight occurs in the target video includes adjusting the luminance of the Y channel value of the corresponding area in the registered source video using an interpolation scheme and replacing the Y channel value of the area in the target video with the adjusted luminance value.

### Claim 10 (depends on 9)

The video restoration apparatus of claim 9, wherein the interpolation scheme uses a formula:
Δ
⁢
p
~
(
i
)
=
1
e
-
s
⁢
(
(
e
-
i
)
⁢
Δ
⁢
p
⁡
(
s
)
+
(
i
-
s
)
⁢
Δ
⁢
p
⁡
(
e
)
)
(wherein, pixel domains p and q denote Y channel values of the source video and the target video, respectively, an i-th pixel value of the source video is p(i), an i-th estimated pixel value of the target video is {tilde over (q)}(i), {tilde over (q)}(i)=p(i)+Δ{tilde over (p)}(i) a start coordinate index of an area with a large difference in Y channel value is s, and an end coordinate index is e, Δp(s)=|q(s)−p(s)|, and Δp(e)=|q(e)−p(e)|).

### Claim 11 (depends on 10)

The video restoration apparatus of claim 10, wherein a formula
Δ
⁢
p
˜
(
i
)
=
1
2
⁢
k
2
⁢
∑
j
⁢
(
Δ
(
j
)
+
Δ
(
j
)
)
(wherein, {tilde over (q)}(i) values of a row and a column are (i)=pr (i)+Δ(i) and (i)=pc (i)+Δ(i), respectively, and k indicates a window size including an area with a large difference in Y channel value) is used to apply the interpolation scheme to each of pixels of all rows and columns of the area in which the light reflection highlight occurs.

### Claim 12 (depends on 7)

The video restoration apparatus of claim 7, further performing:
a process of restoring the area in which the light reflection highlight occurs in the source video by using the target video as a source video for the source video when there is the area in which the light reflection highlight occurs in the source video.
