Dynamic transfer instrument image capture

Patent metadata

Field	Value
Publication number	US-12450708-B2
Application number	US-202318171842-A
Country	US
Kind code	B2
Filing date	Feb 21, 2023
Priority date	Feb 21, 2023
Publication date	Oct 21, 2025
Grant date	Oct 21, 2025

Abstract

Disclosed are systems and methods for generating electronic instruments that implement electronic transfers. The system converts instruments to an electronic format using an imaging source. The image data is processed to determine content elements and segments of the electronic transfer instrument and to extract transfer data. The transfer data is validated to facilitate secure, accurate execution of the electronic transfer.

First claim

What is claimed is:

1. A system for electronic transfer instrument capture comprising a computer including at least one processor and a memory device storing data and executable code that, when executed, causes the at least one processor to:
(a) activate a camera, wherein:
(i) the camera captures a continuous stream of video data that comprises a series of sequential static images, and
(ii) the series of sequential static images comprise (A) a plurality of first side transfer instrument images that depicts a transfer instrument first side, and (B) a plurality of second side transfer instrument images that depicts a transfer instrument second side;

(b) evaluate the first side transfer instrument images by performing operations that
(i) detect edge boundaries for a transfer instrument first side,
(ii) convert the first side transfer instrument image to first side machine encoded content elements to identify typed text and handwritten text on the first side transfer instrument image,
(iii) convert groups of first side machine encoded content elements to first side extracted transfer data elements,
(iv) compare each first side extracted transfer data element to a database of first side expected transfer data elements,
(v) detect a sharpness value for the first side transfer instrument image,
(vi) compare the detected sharpness value against an image sharpness threshold, wherein when the detected sharpness value falls below the image sharpness threshold, the processor ceases evaluation of the first side transfer instrument image and commence evaluation of the next first side transfer instrument image in the sequence of sequential static images,
(vii) detect a signal-to-noise value for the first side transfer instrument image,
(viii) compare the detected signal-to-noise value against an image signal-to-noise threshold wherein when the detected signal-to-noise value falls below the image signal-to-noise threshold, the processor ceases evaluation of the first side transfer instrument image and commence evaluation of the next first side transfer instrument image in the sequence of sequential static images, and
(ix) wherein when, for a given first side transfer instrument image, the edge boundaries are detected and each first side expected transfer data element is matched to at least one first side extracted transfer data element, then (A) the given first side transfer instrument image is stored as an accepted first side transfer instrument image, and (B) the processor does not evaluate any further first side transfer instrument images;

(c) evaluate the second side transfer instrument images by performing operations that
(i) detect edge boundaries for a transfer instrument second side,
(ii) convert second side transfer instrument image to second side machine encoded content elements to identify typed text and handwritten text on the second side transfer instrument image,
(iii) convert groups of second side machine encoded content elements to second side extracted transfer data elements,
(iv) compare each second side extracted transfer data element to a database of second side expected transfer data elements,
(v) detect a sharpness value for the second side transfer instrument image,
(vi) compare the detected sharpness value against an image sharpness threshold, wherein when the detected sharpness value falls below the image sharpness threshold, the processor ceases evaluation of the second side transfer instrument image and commence evaluation of the next second side transfer instrument image in the sequence of sequential static images,
(vii) detect a signal-to-noise value for the second side transfer instrument image,
(viii) compare the detected signal-to-noise value against an image signal-to-noise threshold wherein when the detected signal-to-noise value falls below the image signal-to-noise threshold, the processor ceases evaluation of the second side transfer instrument image and commence evaluation of the next second side transfer instrument image in the sequence of sequential static images, and
(ix) wherein when, for a given second side transfer instrument image, the edge boundaries are detected and each second side expected transfer data element is matched to at least one second side extracted transfer data element, then (A) the given second side transfer instrument image is stored as an accepted second side transfer instrument image, and (B) the processor does not evaluate any further second side transfer instrument images; and

(d) create an electronic transfer instrument comprising the accepted first side transfer instrument image, the accepted second side transfer instrument image, and the first and second side extracted transfer data elements.

2. The system for electronic transfer instrument capture of claim 1 , wherein the first side expected transfer data elements are input by a user to the computer.

3. The system for electronic transfer instrument capture of claim 1 , wherein:
(a) the computer comprises a neural network; and
(b) the neural network is used to read and convert the image data to machine encoded content elements.

4. The system for electronic transfer instrument capture of claim 3 , wherein the neural network has an CNN architecture.

5. The system for electronic transfer instrument capture of claim 3 , wherein the neural network has an recurrent neural network architecture.

6. The system for electronic transfer instrument capture of claim 3 , wherein:
(a) the neural network executes a clustering analysis; and
(b) the neural network comprises a neural network architecture selected from one of a convolutional neural network, a Hopefield network, a Boltzmann Machine, a Helmholtz Machine, a Kohonen Network, a Self-Organizing Map, or a Centroid Neural Network.

7. A system for electronic transfer instrument capture comprising a computer including at least one processor and a memory device storing data and executable code that, when executed, causes the at least one processor to:
(a) activate a camera, wherein:
(i) the camera captures a continuous stream of video data that comprises a series of sequential static images, and
(ii) a plurality of the sequential static images comprise (A) a first side transfer instrument image that depicts a transfer instrument first side, or (B) a second side transfer instrument image that depicts a transfer instrument second side;

(b) evaluate each first side transfer instrument image by performing operations that
(i) convert the first side transfer instrument image to pixels with each pixel having a pixel coordinate position and a brightness value,
(ii) segment the first side transfer instrument image into one or more regions,
(iii) detect edge boundaries of the transfer instrument first side by detecting changes in pixel brightness values above a brightness value threshold,
(iv) processes the first side transfer instrument image to detect one or more dataset arrays identifying clusters of pixels,
(v) feed the one or more dataset arrays to a cluster extraction analysis to identify a first side transfer data element within the first side transfer instrument image, and
(vi) wherein when, for a given first side transfer instrument image, the edge boundaries are detected and at least one transfer data element is matched to a first side expected transfer data element, then (A) the given first side transfer instrument image is stored as an accepted first side transfer instrument image, and (B) the processor does not evaluate any further first side transfer instrument images;

(c) evaluate each second side transfer instrument image by performing operations that
(i) convert the second side transfer instrument image to pixels with each pixel having a pixel coordinate position and a brightness va

Assignees

• Truist Bank

Inventors

• Jayaraman Barath

Classifications

• G06V10/82

  using neural networks · CPC title

• G06T2207/30176

  Document · CPC title

• G06Q20/042

  characterized in that the payment protocol involves at least one cheque · CPC title

• G06N3/0464

  Convolutional networks [CNN, ConvNet] · CPC title

• G06T2207/20084

  Artificial neural networks [ANN] · CPC title