# US 2021/0064871 A1 · Apparatus and method for recognition of text information

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

- Number: US 2021/0064871 A1
- Document ID: US20210064871A1
- Status: Published application
- Kind: A1
- Application number: 16/809,390
- Filed: 2020-03-04
- Published: 2021-03-04
- Earliest claimed: 2019-08-27
- Assignees: LG Electronics Inc.
- Applicants: LG Electronics Inc.
- Inventors: Youngho SOHN; Young Yeon SEO; Changjun YEO
- Web page: https://patentgrep.com/patent/US20210064871A1

## CPC

- G06V 10/44: Local feature extraction by analysis of parts of the pattern, e.g. by detecting edges, contours, loops, corners, strokes or intersections; Connectivity analysis, e.g. of connected components
- G06F 3/0482: Interaction with lists of selectable items, e.g. menus
- G06F 18/2193: {based on specific statistical tests}
- G06V 10/82: using neural networks
- G06V 20/10: Terrestrial scenes (scenes under surveillance with static cameras G06V20/52; scenes perceived from the exterior of a vehicle G06V20/56; scenes perceived from the interior of a vehicle G06V20/59)
- G06V 30/1448: {based on markings or identifiers characterising the document or the area}
- G06V 30/19173: {Classification techniques}
- G06V 30/10: Character recognition
- G06V: IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06F: ELECTRIC DIGITAL DATA PROCESSING (computer systems based on specific computational models G06N)

## Abstract

A text recognition method and apparatus is disclosed. A text recognition method includes detecting the width and the location of a digit through segmentation using a grid having a variable width. According to the present disclosure, it is possible to distinguish an embossing text and a printing text through a 5G network service and a neural network (CNN) performing deep learning, and to recognize a card number through different methods for each card type.

## Claims

### Claim 1

A text recognition method, comprising:
recognizing a card number from a card image by performing vertical segmentation and horizontal segmentation,
wherein the horizontal segmentation comprises detecting a width and a location of a digit of the card number through a segmentation using a grid having a variable width based on a configuration pattern of the card number including a size and a number of digits of the card number.

### Claim 2 (depends on 1)

The text recognition method of claim 1, further comprises performing a Luhn algorithm check, issuer identification number (IIN) check, a check based on multiple image frames, or a CNN mean confidence check for the recognized card number.

### Claim 3

A text recognition method, comprising:
detecting a card area in a card image,
wherein the detecting the card area comprises
recognizing an edge of the card; and
extracting the card area based on the recognized edge,
wherein the recognizing the edge of the card comprises:
analyzing a relationship between an upper edge and a lower edge of the card and a relationship between a left edge and a right edge of the card based on the card image; and
determining the edge of the card based on the relationship analysis.

### Claim 4 (depends on 3)

The text recognition method of claim 3, wherein in the recognizing the edge of the card, an aspect ratio of the card in the card image is used.

### Claim 5

A text recognition method, comprising:
determining a text display method of a card from a card image,
wherein the determining the text display method of the card comprises:
determining at least whether the card image is of a front side or a rear side of the card according to presence of a card magnet in the card image, or whether the card image corresponds to a landscape or portrait orientation of the card based on a text array direction using a margin space calculation between texts; and
predicting the text display method of the card based on the determination.

### Claim 6 (depends on 5)

The text recognition method of claim 5, wherein the determining the text display method of the card further comprises determining the text display method of the card between an embossing method and a printing method.

### Claim 7 (depends on 6)

The text recognition method of claim 6, further comprising recognizing a card number and an expiration date from the card image,
wherein when it is determined that the text display method of the card is the embossing method, the recognizing the card number and the expiration date comprises:
extracting information on at least a card company or an issuer through a card number recognition;
detecting an expiration date type for a particular card company or issuer based on the extracted information and a previously stored database; and
checking the expiration date by using the expiration date type.

### Claim 8 (depends on 6)

The text recognition method of claim 6, further comprising recognizing a card number and an expiration date from the card image,
wherein when it is determined that the text display method of the card is the printing method, the recognizing the card number and the expiration date comprises:
setting a candidate area based on the determination of whether the card image is of the front or rear side or whether the card image corresponds to the portrait or landscape orientation of the card;
recognizing a text in the candidate area through an optical character recognition (OCR); and
predicting the card number by using the recognized text and the expiration date based on a detected slash.

### Claim 9 (depends on 5)

The text recognition method of claim 5, further comprising preprocessing the card image in order to recognize text overlapping an image background.

### Claim 10 (depends on 5)

The text recognition method of claim 5, further comprising:
displaying a menu for selecting a portrait type card or a landscape type card for processing a card image; and
receiving an input card image according to the portrait type card or the landscape type card according to a user selection on the displayed menu.

### Claim 11

A text recognition apparatus, comprising:
a camera;
a display; and
a processor configured to:
capture a card image of a card via the camera and display the card image via the display;
recognizing a card number of the card based on the card image, wherein the card number is recognized by performing vertical segmentation and horizontal segmentation,
wherein the processor is further configured to detect a width and a location of a digit of the card number through vertical segmentation using a grid having a variable width based on a configuration pattern of the card number including a size and a number of digits of the card number.

### Claim 12 (depends on 11)

The text recognition apparatus of claim 11, wherein the processor is further configured to perform a Luhn algorithm check, issuer identification number (IIN) check, a check based on multiple image frames, or a CNN mean confidence check for the recognized card number.

### Claim 13

A text recognition apparatus, comprising:
a camera;
a display; and
a processor configured to:
capture a card image of a card via the camera and display the card image via the display;
detect a card area using the card image by recognizing an edge of the card and extracting the card area based on the recognized edge; and
wherein the processor is configured to recognize the edge of the card by analyzing a relationship between an upper edge and a lower edge of the card and a relationship between a left edge and a right edge of the card based on the card image, and
wherein the processor is further configured to determine the edge of the card based on the relationship analysis.

### Claim 14 (depends on 13)

The text recognition apparatus of claim 13, wherein the processor is further configured to recognize the edge of the card by using an aspect ratio of the card in the card image.

### Claim 15

A text recognition apparatus, comprising:
a camera;
a display; and
a processor configured to:
capture a card image of a card via the camera and display the card image via the display;
determine at least whether the card image is of a front side or a rear side of the card according to presence of a card magnet in the card image, or whether the card image corresponds to a landscape or portrait orientation of the card based on a text array direction using a margin space calculation between texts; and
predict a text display method of a card based on the determination.

### Claim 16 (depends on 15)

The text recognition apparatus of claim 15, wherein the processor is further configured to determine the text display method of the card between an embossing method and a printing method.

### Claim 17 (depends on 16)

The text recognition apparatus of claim 16, wherein the processor is further configured to:
extract information on at least a card company or an issuer through a card number recognition;
detect an expiration date type for a particular card company or issuer based on the extracted information and a previously stored database;
check an expiration date by using a database on an expiration date type for each card company and issuer previously prepared based on the recognized card number when the text display method of the card is determined to be the embossing method.

### Claim 18 (depends on 16)

The text recognition apparatus of claim 16, wherein the processor is further configured to:
set a candidate area based on the determination of whether the card image is of the front or rear side or whether the card image corresponds to the portrait or landscape orientation of the card;
recognize a text in the candidate area through an optical character recognition (OCR); and
predict a card number by using the recognized text and an expiration date based on a detected slash.

### Claim 19 (depends on 15)

The text recognition apparatus of claim 15, wherein the processor is further configured to perform a card image preprocessing in order to recognize text overlapping an image background.

### Claim 20

A text recognition apparatus, comprising:
a camera;
a display; and
a processor configured to:
display a menu via the display for selecting a portrait type card or a landscape type card for processing the card image; and
receiving an input card image captured via the camera according to the portrait type card or the landscape type card according to a user selection on the displayed menu.
