# US 2026/0148392 A1 · Estimation of the movement of an image position

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

- Number: US 2026/0148392 A1
- Document ID: US20260148392A1
- Status: Published application
- Kind: A1
- Application number: 19/177,872
- Filed: 2025-04-14
- Published: 2026-05-28
- Earliest claimed: 2019-03-01
- Assignees: Hochschule Rheinmain University of Applied Sciences Wiesbaden Ruesselsheim
- Applicants: Hochschule Rheinmain University of Applied Sciences Wiesbaden Ruesselsheim
- Inventors: Matthias NARROSCHKE; Andreas KAH
- Web page: https://patentgrep.com/patent/US20260148392A1

## CPC

- G06T 7/269: using gradient-based methods
- G06T 7/207: for motion estimation over a hierarchy of resolutions (multi-resolution motion estimation or hierarchical motion estimation for coding, decoding, compressing or decompressing digital video signals H04N19/53)
- G06T 7/248: {involving reference images or patches}
- G06T: IMAGE DATA PROCESSING OR GENERATION, IN GENERAL

## Abstract

The invention relates to a computer-implemented method for estimating the movement of an image position of an image feature between a first image and a second image. The method includes: Determining a set of image windows in the first image, each image window having the image feature; Identifying at least a portion of the set of image windows in the second image; For the portion of the set of image windows, evaluating the quality of an assignment of each image window in the first image to the given image window in the second image, and selecting one of the image windows on the basis of the evaluation; and estimating the movement of the image position of the image feature using the selected image window. In the second image, the selected window is identified by iteration over more resolutions than the windows of the at least one portion.

## Claims

### Claim 1

A computer-implemented method for estimating the movement of an image position of an image feature between a first image and a second image, the method comprising:
determining a set of image windows in the first image, each image window having the image feature,
identifying at least a portion of the set of image windows in the second image,
for the portion of the set of image windows, evaluating the quality of an assignment of each image window in the first image to the given image window in the second image, and selecting one of the image windows on the basis of the evaluation,
estimating the movement of the image position of the image feature using the selected image window,
wherein
for each image window of the at least one portion of the set of image windows, the identification of the image window in the second image takes place iteratively, with a first number of different image resolutions of the first and second images, wherein, starting from the lowest image resolution, the result of the identification of the image window is used for identifying the image window with a higher image resolution, and
the identification of the selected image window in the second image takes place iteratively, with a second number of different image resolutions of the first and second image, wherein the result of the identification of the selected image window is used, starting from the lowest image resolution, to identify the selected image window at a higher image resolution,
wherein the first number is less than the second number.

### Claim 2 (depends on 1)

The method according to claim 1, wherein the set of image windows surrounds the image position in the first image point-symmetrically with respect to the image position.

### Claim 3 (depends on 1)

The method according to claim 1, wherein the set of image windows comprises fewer than 10 image windows, in particular 9, 5 or 3, or fewer than 3 image windows.

### Claim 4 (depends on 1)

The method according to claim 1, wherein the set of image windows comprises an initial image window, and the method comprises a detection of an edge in the initial image window, wherein the determination of the image windows contained in addition to the initial image window in the set of image windows takes place on the basis of the detected edge.

### Claim 5 (depends on 4)

The method according to claim 4, wherein the image windows of the set of image windows are arranged one behind the other perpendicular to the detected edge.

### Claim 6 (depends on 5)

The method according to claim 5, wherein the initial image window is not taken into account in obtaining the portion of the set of image windows.

### Claim 7 (depends on 1)

The method according to claim 1, wherein, in obtaining the portion of the set of image windows, a randomly selected set of image windows is not taken into account.

### Claim 8 (depends on 1)

The method according to claim 1, wherein the identification of the at least on portion of the set of image windows in the second image takes place using an optical flow determination method.

### Claim 9 (depends on 8)

The method according to claim 8, the optical flow determination method being the Lucas-Kanade method or the Horn-Schunck method.

### Claim 10 (depends on 8)

The method according to claim 8, wherein the detection of the edge is based on the structure tensor of the Harris method or the Lucas-Kanade method.

### Claim 11 (depends on 10)

The method according to claim 10, wherein the determination of the image windows contained in addition to the initial image window in the set of image windows on the basis of the detected edge comprises: Determining a first direction based on the eigenvector of the structure tensor which corresponds to the greatest eigenvalue, and determination of a second direction perpendicular to the first direction, the image windows contained in the set of image windows in addition to the initial image window being arranged along the second direction.

### Claim 12 (depends on 1)

The method according to claim 1, wherein the movement is estimated on the basis of the difference in position of the selected image window between the first image and the second image.

### Claim 13 (depends on 12)

The method according to claim 12, wherein the estimation of the movement comprises an identification of the selected image window in the second image, wherein
for each image window of the at least one portion of the set of image windows, the identification of the image window in the second image takes place iteratively, with a first number of iteration steps for different positions of each image window in the second image, and
the selected image window in the second image is identified iteratively with a second number of iteration steps for different positions of the selected image window in the second image,
wherein the second number is greater than the first number.

### Claim 14 (depends on 12)

The method according to claim 12, wherein the estimation of the movement is carried out using an optical flow determination method.

### Claim 15 (depends on 1)

The method according to claim 1, wherein
the portion of the set of image windows in the second image has first image sizes, and
to estimate the movement of the image position of the image feature, the selected image window is enlarged compared to the associated first image size.

### Claim 16 (depends on 1)

The method according to claim 1, wherein, with respect to a given image window of the portion of the set of image windows in the first image, the evaluation of the quality is selected from one of the following criteria:
starting from the identified image window in the second image, identification of an associated image window in the first image, wherein the evaluation of the quality is a measure of the congruence of the identified associated image window in the first image with the image window of the set of image windows in the first image, optionally with an identification of the associated image window in the first image using an optical flow determination method;
determination of a degree of scattering of the image content of the image window in the first image with respect to the identified image window in the second image, the evaluation of the quality being based on the degree of scattering;
determination of the Lagrangian costs of a degree of scatter of the image content of the image window in the first image with respect to the identified image window in the second image, and determination of a further measure, the further measure being based on one of the following variants:
the gradient of the image content of the image window in the first image or in the second image;
the structure tensor of the Harris method or the Lucas-Kanade method;
a cornerness measure of the image content of the image window in the first image or in the second image.

### Claim 17 (depends on 16)

The method according to claim 16, wherein the scattering comprises a mean or absolute square error.

### Claim 18 (depends on 16)

The method according to claim 16, wherein
the identified image window in the second image is displaced by a first vector compared to the given image window in the first image,
the associated image window in the first image is displaced by a second vector compared to the identified image window in the second image, and
the evaluation of the quality takes into account a comparison of the first and second vector.

### Claim 19 (depends on 1)

A computer program product having instructions that can be executed by a processor for carrying out the method according to claim 1.

### Claim 20

A device for estimating the movement of an image position of an image feature between a first image and a second image, the apparatus comprising a processor and a memory with instructions, wherein execution of the instructions by the processor results in the device:
determining a set of image windows in the first image, each image window having the image feature,
identifying at least a portion of the set of image windows in the second image,
for the portion of the set of image windows, evaluating the quality of an assignment of each image window in the first image to the given image window in the second image, and selecting one of the image windows on the basis of the evaluation,
estimating the movement of the image position of the image feature using the selected image window,
wherein
for each image window of the at least one portion of the set of image windows, the identification of the image window in the second image takes place iteratively, with a first number of different image resolutions of the first and second images, wherein, starting from the lowest image resolution, the result of the identification of the image window is used for identifying the image window with a higher image resolution, and
the identification of the selected image window in the second image takes place iteratively, with a second number of different image resolutions of the first and second image, wherein the result of the identification of the selected image window is used, starting from the lowest image resolution, to identify the selected image window at a higher image resolution,
wherein the first number is less than the second number.
