# US 2025/0156994 A1 · Image Processing Method, Apparatus, and System

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

- Number: US 2025/0156994 A1
- Document ID: US20250156994A1
- Status: Published application
- Kind: A1
- Application number: 19/022,660
- Filed: 2025-01-15
- Published: 2025-05-15
- Earliest claimed: 2022-08-22
- Assignees: Huawei Technologies Co., Ltd.
- Applicants: Huawei Technologies Co., Ltd.
- Inventors: Xin Wang; Jing Huang; Xiaoxu Zheng; Michael Bi Mi
- Web page: https://patentgrep.com/patent/US20250156994A1

## CPC

- G06T 3/4046: using neural networks
- G06T 3/4069: by subpixel displacements
- G06T 3/40: Scaling of whole images or parts thereof, e.g. expanding or contracting
- G06T 3/4053: based on super-resolution, i.e. the output image resolution being higher than the sensor resolution
- H04N 19/132: Sampling, masking or truncation of coding units, e.g. adaptive resampling, frame skipping, frame interpolation or high-frequency transform coefficient masking
- H04N 19/172: the region being a picture, frame or field
- H04N 19/59: involving spatial sub-sampling or interpolation, e.g. alteration of picture size or resolution
- G06T: IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- H04N: PICTORIAL COMMUNICATION, e.g. TELEVISION

## Abstract

An image processing method includes obtaining image data, where the image data includes a plurality of consecutive first image frames; sampling points in different phases for different first image frames in the consecutive first image frames, so that at least two adjacent second image frames in a plurality of consecutive second image frames that are obtained through downsampling have different sampling points in same pixel modules. The plurality of consecutive second image frames include pixel information of different locations of a same object; and performing super-resolution processing so that the pixel information that is of different locations of the same object and that is in the plurality of consecutive second image frames is used to restore the object.

## Claims

### Claim 1

An image processing method, comprising:
obtaining image data, wherein the image data comprises a plurality of first consecutive image frames; and
separately performing downsampling processing on the first consecutive image frames using a downsampling network to obtain a plurality of second consecutive image frames,
wherein the second consecutive image frames are in a one-to-one correspondence with the first consecutive image frames, and
wherein at least two adjacent second image frames in the second consecutive image frames have different sampling points in the same pixel modules of the at least two adjacent second frames.

### Claim 2 (depends on 1)

The image processing method of claim 1, wherein separately performing the downsampling processing on the first consecutive image frames comprises separately performing downsampling processing on the of first consecutive image frames based on at least two preset phases, wherein each of the first consecutive image frames corresponds to one of the at least two preset phases, wherein two adjacent first image frames in the first image frames correspond to different preset phases of the at least two preset phases.

### Claim 3 (depends on 1)

The image processing method of claim 1, wherein separately performing the downsampling processing on the first consecutive image frames to obtain the second consecutive image frames comprises:
performing downsampling processing on each of the first consecutive image frames based on at least two preset phases to obtain at least two candidate image frames; and
selecting candidate image frames from the at least two candidate image frames to obtain the second consecutive image frames, wherein preset phases used for downsampling two corresponding candidate image frames for two adjacent first image frames of the first consecutive image frames are different.

### Claim 4 (depends on 1)

The image processing method of claim 1, wherein each of the sampling points is a pixel or a sub-pixel.

### Claim 5

(canceled)

### Claim 6 (depends on 1)

The image processing method of claim 1, wherein further comprising:
performing downsampling training on a plurality of training image frames to obtain a plurality of sampling image frames, wherein at least two adjacent sampling image frames in the sampling image frames have different sampling points in the same pixel modules;
performing super-resolution training on the sampling image frames to obtain a plurality of training restoration image frames; and
determining the downsampling network based on the training restoration image frames and the training image frames.

### Claim 7 (depends on 1)

The image processing method of claim 1, further comprising encoding the second consecutive image frames to obtain image encoding data.

### Claim 8

An image processing method, comprising:
obtaining a plurality of second consecutive image frames, based on separately downsampling a plurality of consecutive first image frames, wherein the second consecutive image frames are in a one-to-one correspondence with the first consecutive image frames, and wherein at least two adjacent second image frames in the second consecutive image frames have different sampling points in the same pixel modules of the at least two adjacent second image frames; and
performing super-resolution processing on the second consecutive image frames to obtain a plurality of third consecutive image frames, wherein the second consecutive image frames are in a one-to-one correspondence with the third consecutive image frames.

### Claim 9 (depends on 8)

The image processing method of claim 8, wherein obtaining the second consecutive image frames comprises:
obtaining image encoding data; and
decoding the image encoding data to obtain the second consecutive image frames.

### Claim 10 (depends on 9)

The image processing method of claim 9, wherein performing the super-resolution processing on the second consecutive image frames comprises performing super-resolution processing on the second consecutive image frames using a super-resolution network to obtain the third consecutive image frames.

### Claim 11 (depends on 10)

The image processing method of claim 10, further comprising:
performing downsampling processing on a plurality of training image frames using a downsampling network to obtain a plurality of sampling image frames;
obtaining a plurality of degraded sampling image frames by performing encoding and decoding processing on the sampling image frames; and
performing super-resolution training on the degraded sampling image frames to obtain the super-resolution network.

### Claim 12 (depends on 8)

The image processing method of claim 8, wherein each of the sampling points is a pixel or a sub-pixel.

### Claim 13

An image processing apparatus, comprising:
a memory storing instructions; and
at least one processor coupled to the memory and configured to execute the instructions to cause the image processing apparatus to;
obtain image data, wherein the image data comprises a plurality of first consecutive image frames; and
separately perform downsampling processing on the first consecutive image frames using a downsampling network to obtain a plurality of second consecutive second image frames,
wherein the second consecutive image frames are in a one-to-one correspondence with the first consecutive image frames, and
wherein at least two adjacent second image frames in the second consecutive image frames have different sampling points in same pixel modules of the at least two adjacent second frames.

### Claim 14 (depends on 13)

The image processing apparatus of claim 13, wherein the instructions, when executed by the at least one processor, further cause the image processing apparatus to separately perform downsampling processing on the first consecutive image frames based on at least two preset phases, wherein each of the first consecutive image frames corresponds to one of the at least two preset phases, wherein two adjacent first image frames correspond to different preset phases of the at least two preset phases.

### Claim 15 (depends on 13)

The image processing apparatus of claim 13, wherein the instructions, when executed by the at least one processor further cause the image processing apparatus to:
perform downsampling processing on each of first consecutive image frames based on at least two preset phases, to obtain at least two candidate image frames; and
select candidate image frames from the at least two candidate image frames to obtain the second consecutive image frames, wherein preset phases used for downsampling two corresponding candidate image frames for two adjacent first image frames of the first consecutive image frames are different.

### Claim 16 (depends on 13)

The image processing apparatus of claim 13, wherein each of the sampling points is a pixel or a sub-pixel.

### Claim 17

(canceled)

### Claim 18 (depends on 13)

The image processing apparatus of claim 13, wherein the instructions, when executed by the at least one processor further cause the image processing apparatus to:
perform downsampling training on a plurality of training image frames to obtain a plurality of sampling image frames, wherein at least two adjacent sampling image frames in the sampling image frames have different sampling points in the same pixel modules;
perform super-resolution training on the sampling image frames to obtain a plurality of training restoration image frames; and
determine the downsampling network based on the training restoration image frames and the training image frames.

### Claim 19 (depends on 13)

The image processing apparatus of claim 13, wherein the instructions, when executed by the at least one processor, further cause the image processing apparatus to encode the second consecutive image frames to obtain image encoding data.

### Claim 20 (depends on 13)

The image processing apparatus of claim 13, wherein the instructions, when executed by the at least one processor, further cause the image processing apparatus to obtain the downsampling network based on plurality of training restoration image frames and a plurality of training image frames.

### Claim 21 (depends on 1)

The image processing method of claim 1, further comprising obtaining the downsampling network based on a plurality of training restoration image frames and a plurality of training image frames.

### Claim 22 (depends on 11)

The image processing method of claim 11, wherein at least two adjacent sampling image frames in the sampling image frames have different sampling points in the same pixel modules of the at least two adjacent second frames.
