# US 2023/0367854 A1 · Creating feature based image recognition subclasses for identity verification

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

- Number: US 2023/0367854 A1
- Document ID: US20230367854A1
- Status: Published application
- Kind: A1
- Application number: 17/662,695
- Filed: 2022-05-10
- Published: 2023-11-16
- Earliest claimed: 2022-05-10
- Assignees: Bank of America Corporation
- Applicants: Bank of America Corporation
- Inventors: Maharaj Mukherjee
- Web page: https://patentgrep.com/patent/US20230367854A1

## CPC

- G06F 21/32: using biometric data, e.g. fingerprints, iris scans or voiceprints
- G06N 3/08: Learning methods
- G06V 10/431: {Frequency domain transformation; Autocorrelation}
- G06V 10/435: {Computation of moments}
- G06V 10/454: {Integrating the filters into a hierarchical structure, e.g. convolutional neural networks [CNN]}
- G06V 10/764: using classification, e.g. of video objects
- G06V 10/7747: {Organisation of the process, e.g. bagging or boosting}
- G06V 10/82: using neural networks
- G06V 40/1365: {Matching; Classification}
- G06V 40/172: {Classification, e.g. identification}
- G06V 40/197: {Matching; Classification}
- G06V 40/25: {Recognition of walking or running movements, e.g. gait recognition}
- G06V 40/70: Multimodal biometrics, e.g. combining information from different biometric modalities
- G06N 3/0464: Convolutional networks [CNN, ConvNet]
- G06N 3/09: Supervised learning
- G06F: ELECTRIC DIGITAL DATA PROCESSING (computer systems based on specific computational models G06N)
- G06N: COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06V: IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING

## Abstract

Aspects of the disclosure relate to user authentication. A computing platform may receive a plurality of facial scans of an individual. The computing platform may train, using the plurality of facial scans, a convolutional neural network (CNN) to identify the individual, based on a first facial scan of the individual, using subclasses of the CNN. The computing platform may receive an authorization request including the first facial scan of the individual. The computing platform may input the first facial scan into the CNN, which may cause the CNN to identify the individual. Based on successful identification of the individual, the computing platform may grant requested access to the individual. The computing platform may update, using the first facial scan, the CNN.

## Claims

### Claim 1

A computing platform comprising:
at least one processor;
a communication interface communicatively coupled to the at least one processor; and
memory storing computer-readable instructions that, when executed by the at least one processor, cause the computing platform to:
receive a plurality of facial scans of an individual;
train, using the plurality of facial scans, a convolutional neural network (CNN) to identify the individual based on a first facial scan of the individual, wherein training the CNN comprises:
converting the plurality of facial scans to a frequency domain,
identifying a center of gravity for each of the plurality of facial scans, and
generating, based on distances between the centers of gravity, one or more subclasses for the plurality of facial scans, wherein each subclass is defined by a center of gravity range limited by a minimum center of gravity value and a maximum center of gravity value;
receive an authorization request including the first facial scan of the individual;
input the first facial scan into the CNN, wherein inputting the first facial scan into the CNN causes the CNN to:
convert the first facial scan to the frequency domain,
identify a center of gravity for the first facial scan,
compare the center of gravity for the first facial scan to center of gravity ranges to identify a corresponding subclass, and
identify, based on the corresponding subclass, the individual;
based on successful identification of the individual, grant requested access to the individual; and
update, using the first facial scan, the CNN.

### Claim 2 (depends on 1)

The computing platform of claim 1, wherein the plurality of facial scans comprise images of the individual throughout a lifetime of the individual.

### Claim 3 (depends on 1)

The computing platform of claim 1, wherein training the CNN further comprises training the CNN based on a plurality of fingerprint patterns of the individual, and wherein training the CNN comprises:
converting the plurality of fingerprint patterns to the frequency domain,
identifying a center of gravity for each of the plurality of fingerprint patterns, and
generating, based on distances between the centers of gravity of the plurality of fingerprint patterns, one or more subclasses for the plurality of fingerprint patterns.

### Claim 4 (depends on 3)

The computing platform of claim 3, wherein the memory stores additional computer-readable instructions that, when executed by the at least one processor, cause the computing platform to:
receive a first fingerprint pattern;
input the first fingerprint pattern into the CNN, wherein inputting the first fingerprint pattern into the CNN causes the CNN to:
convert the first fingerprint pattern to the frequency domain;
identify a center of gravity for the first fingerprint pattern;
identify, based on the center of gravity for the first fingerprint pattern, a corresponding fingerprint pattern subclass; and
identify, based on the corresponding fingerprint pattern subclass, the individual.

### Claim 5 (depends on 1)

The computing platform of claim 1, wherein training the CNN further comprises training the CNN based on a plurality of iris images of the individual, and wherein training the CNN comprises:
converting the plurality of iris images to the frequency domain,
identifying a center of gravity for each of the plurality of iris images, and
generating, based on distances between the centers of gravity of the plurality of iris images, one or more subclasses for the plurality of iris images.

### Claim 6 (depends on 5)

The computing platform of claim 5, wherein the memory stores additional computer-readable instructions that, when executed by the at least one processor, cause the computing platform to:
receive a first iris image;
input the first iris image into the CNN, wherein inputting the first iris image into the CNN causes the CNN to:
convert the first iris image to the frequency domain;
identify a center of gravity for the first iris image;
identify, based on the center of gravity for the first iris image, a corresponding iris subclass; and
identify, based on the corresponding iris subclass, the individual.

### Claim 7 (depends on 1)

The computing platform of claim 1, wherein training the CNN further comprises training the CNN based on a plurality of gait images of the individual, and wherein training the CNN comprises:
converting the plurality of gait images to the frequency domain,
identifying a center of gravity for each of the plurality of gait images, and
generating, based on distances between the centers of gravity of the plurality of gait images, one or more subclasses for the plurality of gait images.

### Claim 8 (depends on 7)

The computing platform of claim 7, wherein the memory stores additional computer-readable instructions that, when executed by the at least one processor, cause the computing platform to:
receive a first gait image;
input the first gait image into the CNN, wherein inputting the first gait image into the CNN causes the CNN to:
convert the first gait image to the frequency domain,
identify a center of gravity for the first gait image,
identify, based on the center of gravity for the first gait image, a corresponding gait subclass, and
identify, based on the corresponding gait subclass, the individual.

### Claim 9 (depends on 1)

The computing platform of claim 1, wherein identifying the individual further comprises identifying, based on analysis of additional biometric information using the CNN, the individual.

### Claim 10 (depends on 1)

The computing platform of claim 1, wherein updating the CNN comprises one or more of:
adding a new subclass based on the center of gravity for the first facial scan, or
merging the subclasses based on the center of gravity for the first facial scan.

### Claim 11

A method comprising:
at a computing platform comprising at least one processor, a communication interface, and memory:
receiving a plurality of facial scans of an individual;
training, using the plurality of facial scans, a convolutional neural network (CNN) to identify the individual based on a first facial scan of the individual, wherein training the CNN comprises:
converting the plurality of facial scans to a frequency domain,
identifying a center of gravity for each of the plurality of facial scans, and
generating, based on distances between the centers of gravity, one or more subclasses for the plurality of facial scans, wherein each subclass is defined by a center of gravity range limited by a minimum center of gravity value and a maximum center of gravity value;
receiving an authorization request including the first facial scan of the individual;
inputting the first facial scan into the CNN, wherein inputting the first facial scan into the CNN causes the CNN to:
converting the first facial scan to the frequency domain,
identifying a center of gravity for the first facial scan,
comparing the center of gravity for the first facial scan to center of gravity ranges to identify a corresponding subclass, and
identifying, based on the corresponding subclass, the individual;
based on successful identification of the individual, granting requested access to the individual; and
updating, using the first facial scan, the CNN.

### Claim 12 (depends on 11)

The method of claim 11, wherein the plurality of facial scans comprise images of the individual throughout a lifetime of the individual.

### Claim 13 (depends on 11)

The method of claim 11, wherein training the CNN further comprises training the CNN based on a plurality of fingerprint patterns of the individual, and wherein training the CNN comprises:
converting the plurality of fingerprint patterns to the frequency domain,
identifying a center of gravity for each of the plurality of fingerprint patterns, and
generating, based on distances between the centers of gravity of the plurality of fingerprint patterns, one or more subclasses for the plurality of fingerprint patterns.

### Claim 14 (depends on 13)

The method of claim 13, further comprising:
receiving a first fingerprint pattern;
inputting the first fingerprint pattern into the CNN, wherein inputting the first fingerprint pattern into the CNN causes the CNN to:
convert the first fingerprint pattern to the frequency domain;
identify a center of gravity for the first fingerprint pattern;
identify, based on the center of gravity for the first fingerprint pattern, a corresponding fingerprint pattern subclass; and
identify, based on the corresponding fingerprint pattern subclass, the individual.

### Claim 15 (depends on 11)

The method of claim 11, wherein training the CNN further comprises training the CNN based on a plurality of iris images of the individual, and wherein training the CNN comprises:
converting the plurality of iris images to the frequency domain,
identifying a center of gravity for each of the plurality of iris images, and
generating, based on distances between the centers of gravity of the plurality of iris images, one or more subclasses for the plurality of iris images.

### Claim 16 (depends on 15)

The method of claim 15, further comprising:
receiving a first iris image;
inputting the first iris image into the CNN, wherein inputting the first iris image into the CNN causes the CNN to:
convert the first iris image to the frequency domain;
identify a center of gravity for the first iris image;
identify, based on the center of gravity for the first iris image, a corresponding iris subclass; and
identify, based on the corresponding iris subclass, the individual.

### Claim 17 (depends on 11)

The method of claim 11, wherein training the CNN further comprises training the CNN based on a plurality of gait images of the individual, and wherein training the CNN comprises:
converting the plurality of gait images to the frequency domain,
identifying a center of gravity for each of the plurality of gait images, and
generating, based on distances between the centers of gravity of the plurality of gait images, one or more subclasses for the plurality of gait images.

### Claim 18 (depends on 17)

The method of claim 17, further comprising:
receiving a first gait image;
inputting the first gait image into the CNN, wherein inputting the first gait image into the CNN causes the CNN to:
convert the first gait image to the frequency domain,
identify a center of gravity for the first gait image,
identify, based on the center of gravity for the first gait image, a corresponding gait subclass, and
identify, based on the corresponding gait subclass, the individual.

### Claim 19 (depends on 11)

The method of claim 11, wherein identifying the individual further comprises identifying, based on analysis of additional biometric information using the CNN, the individual.

### Claim 20

One or more non-transitory computer-readable media storing instructions that, when executed by a computing platform comprising at least one processor, a communication interface, and memory, cause the computing platform to:
receive a plurality of facial scans of an individual;
train, using the plurality of facial scans, a convolutional neural network (CNN) to identify the individual based on a first facial scan of the individual, wherein training the CNN comprises:
converting the plurality of facial scans to a frequency domain,
identifying a center of gravity for each of the plurality of facial scans, and
generating, based on distances between the centers of gravity, one or more subclasses for the plurality of facial scans, wherein each subclass is defined by a center of gravity range limited by a minimum center of gravity value and a maximum center of gravity value;
receive an authorization request including the first facial scan of the individual;
input the first facial scan into the CNN, wherein inputting the first facial scan into the CNN causes the CNN to:
convert the first facial scan to the frequency domain,
identify a center of gravity for the first facial scan,
compare the center of gravity for the first facial scan to center of gravity ranges to identify a corresponding subclass, and
identify, based on the corresponding subclass, the individual;
based on successful identification of the individual, grant requested access to the individual; and
update, using the first facial scan, the CNN.
