# US 2023/0267293 A1 · Locating code image zones in an image of a code bearing object

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

- Number: US 2023/0267293 A1
- Document ID: US20230267293A1
- Status: Published application
- Kind: A1
- Application number: 18/108,243
- Filed: 2023-02-10
- Published: 2023-08-24
- Earliest claimed: 2022-02-21
- Assignees: Sick AG
- Applicants: Sick AG
- Inventors: Pascal SCHÜLER; Christofer ROSER; Thorsten FALK; Sascha BURGHARDT; Romain MÜLLER
- Web page: https://patentgrep.com/patent/US20230267293A1

## CPC

- G06K 7/1443: {locating of the code in an image}
- G06K 7/1482: {using fuzzy logic or natural solvers, such as neural networks, genetic algorithms and simulated annealing}
- G06K 7/1413: {1D bar codes}
- G06K 7/1417: {2D bar codes}
- G06K 7/1491: {the method including a reconstruction step, e.g. stitching two pieces of bar code together to derive the full bar code}
- G06K: GRAPHICAL DATA READING (image or video recognition or understanding G06V); PRESENTATION OF DATA; RECORD CARRIERS; HANDLING RECORD CARRIERS

## Abstract

A method of locating code image zones in an output image of a code bearing object (14), wherein first candidates for code image zones are determined in a first segmentation process using a process of classical image processing without machine learning and second candidates for code image zones are determined in a second segmentation process using machine learning, with the first and second candidates being fused to locate the code image zones.

## Claims

### Claim 1

A computer implemented method of locating code image zones in an output image of a code bearing object, wherein first candidates for code image zones are determined in a first segmentation process using a process of classical image processing without machine learning, wherein second candidates for code image zones are determined in a second segmentation process using machine learning: and wherein the first and second candidates are fused to locate the code image zones.

### Claim 2 (depends on 1)

The method in accordance with claim 1,
wherein the first segmentation process generates a first result map and/or the second segmentation process generates a second result map, with a result map being an image of lower resolution than the output image whose pixels comprise information on whether a code image zone has been recognized at the location of the pixel.

### Claim 3 (depends on 1)

The method in accordance with claim 1,
wherein, subsequently to the locating of the candidates place, a fine segmentation takes place in which the code image zones are delineated more finely.

### Claim 4 (depends on 1)

The method in accordance with claim 1,
wherein, subsequently to the locating of the candidates place, a fine segmentation takes place in which the code image zones are delineated more finely in the resolution of the output image.

### Claim 5 (depends on 1)

The method in accordance with claim 1,
wherein a code image zone is deemed to have been located in the output image where a first candidate and a second candidate have been determined or alternatively where a first or second candidate has been determined.

### Claim 6 (depends on 1)

The method in accordance with claim 1,
wherein the first candidates are complemented to form connected code image zones; and wherein only small first candidates are maintained that are smaller than a threshold size.

### Claim 7 (depends on 6)

The method in accordance with claim 6,
wherein the small first candidates are additionally considered as code image zones.

### Claim 8 (depends on 6)

The method in accordance with claim 6,
wherein the first candidates are complemented to form connected code image zones by means of connected component processes.

### Claim 9 (depends on 1)

The method in accordance with claim 1,
wherein the second candidates are complemented to form connected image zones; and wherein only exclusive second candidates are maintained whose position does not coincide with a position of an already located code image zone.

### Claim 10 (depends on 9)

The method in accordance with claim 9,
wherein the exclusive second candidates are additionally considered as code image zones.

### Claim 11 (depends on 9)

The method in accordance with claim 9,
wherein the second candidates are complemented to form connected image zones by means of connected component processes.

### Claim 12 (depends on 1)

The method in accordance with claim 1,
wherein a scoring is determined with respect to the first candidates, the second candidates, and/or the code image zones that indicates how reliably an optical code in the output image has been recognized at the represented position.

### Claim 13 (depends on 1)

The method in accordance with claim 1,
wherein first candidates for 1D codes, first candidates for 2D codes, second candidates for 1D codes, and/or second candidates for 2D codes are determined.

### Claim 14 (depends on 1)

The method in accordance with claim 1,
wherein first candidates that are not simultaneously second candidates are considered as a text image zone within the output image.

### Claim 15 (depends on 1)

The method in accordance with claim 1,
wherein the first segmentation process determines first candidates in an output image divided into tiles.

### Claim 16 (depends on 1)

The method in accordance with claim 1,
wherein the first segmentation process has a contrast determination in which first candidates have to have a minimum contrast.

### Claim 17 (depends on 1)

The method in accordance with claim 1,
wherein, in the first segmentation process, brightness edges are counted transversely to one another along two lines to determine a preferred direction with reference to the respective number of brightness edges and a first candidate for a 1D code is only recognized in the case of a preferred direction.

### Claim 18 (depends on 17)

The method in accordance with claim 17,
wherein, otherwise, a first candidate for a 2D code is recognized.

### Claim 19 (depends on 1)

The method in accordance with claim 1,
wherein the second segmentation process has a neural network.

### Claim 20 (depends on 19)

The method in accordance with claim 19,
wherein the neural network is a convolution neural network.

### Claim 21 (depends on 19)

The method in accordance with claim 19,
wherein the neural network is trained by means of supervised learning using example images that are in particular evaluated using the results of a segmentation and/or decoding process without processes of machine learning.

### Claim 22

An optoelectronic code reader having at least one light receiver element for generating image data from received light and having a control and evaluation unit in which a method of locating code image zones in an output image of a code bearing object is implemented, in which first candidates for code image zones are determined in a first segmentation process using a process of classical image processing without machine learning,
wherein, in the method implemented in the control and evaluation unit, second candidates for code image zones are determined in a second segmentation process using machine learning; and the first and second candidates are fused to locate the code image zones.

### Claim 23 (depends on 1)

An optoelectronic code reader having at least one light receiver element for generating image data from received light and having a control and evaluation unit in which a method of locating code image zones in an output image of a code bearing object is implemented, in which first candidates for code image zones are determined in a first segmentation process using a process of classical image processing without machine learning,
wherein, in the method implemented in the control and evaluation unit, second candidates for code image zones are determined in a second segmentation process using machine learning; and the first and second candidates are fused to locate the code image zones,
wherein the control and evaluation unit is configured to carry out the computer implemented method of locating code image zones in an output image of a code bearing object in accordance with claim 1.
