Automated Image Validation System for Fraud Detection

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Businesses face inefficiencies and costs due to the time-consuming and resource-intensive manual verification of images provided by customers for support or refunds, as fraudulent modifications can lead to ineligible concessions, necessitating a more automated and accurate method to assess image authenticity.

Innovation Solution

An automated system performs a multi-factor analysis on images using modification indicators such as pixel density changes, metadata analysis, and comparisons with stored images to determine the likelihood of tampering, providing an indication of image modification that can be used by customer support agents to make informed decisions.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If manual verification of images is performed by support agents, then image authenticity can be confirmed, but time consumption and operational costs increase significantly

Engineering Contradiction:
Improveimage authenticity verificationVSAvoidverification time
Core Design Contradiction:
ReliabilityVSLoss of time

Solution Approach 1:

The patent replaces manual mechanical verification by support agents with an automated image forensics system that uses digital signal processing and machine learning algorithms to analyze image authenticity, thereby eliminating time-consuming human inspection while maintaining verification reliability

Inventive Principle:
Principle #28Mechanics substitution (Replace mechanical system)

Solution Approach 2:

The patent introduces an intermediary image forensics analysis system that acts as a mediator between image submission and support decision-making, automatically detecting tampering through multi-factor analysis including metadata examination, pixel-level anomaly detection, and consistency verification across image portions

Inventive Principle:
Principle #24Intermediary (Mediator)

2Measurement precision

If sophisticated image forensics processes are implemented, then detection accuracy improves, but computing resources and processing time increase

Engineering Contradiction:
Improvetampering detection accuracyVSAvoidcomputing resource consumption
Core Design Contradiction:
Measurement precisionVSUse of energy by moving object

Solution Approach 1:

The patent segments the image analysis process into multiple independent modules including metadata analysis, pixel density examination, portion-based consistency checking, and machine learning-based anomaly detection, allowing selective execution of analysis steps based on risk assessment to reduce overall computing resource consumption while maintaining high detection accuracy

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent implements a tiered analysis approach where not all forensics techniques are applied to every image uniformly; instead, the system performs basic analysis on all images and applies more computationally intensive techniques only when initial indicators suggest potential tampering, optimizing the balance between detection accuracy and resource usage

Inventive Principle:
Principle #16Partial or excessive action

3Ease of operation

If manual verification processes are used, then customer service quality can be maintained, but operational costs for businesses increase

Engineering Contradiction:
Improvecustomer service qualityVSAvoidoperational costs
Core Design Contradiction:
Ease of operationVSLoss of energy

Solution Approach 1:

The patent enables self-service image verification where the automated forensics system independently performs comprehensive authenticity checks without requiring human intervention, freeing support agents to focus on customer interaction while the system handles verification autonomously, thereby maintaining service quality while reducing operational costs

Inventive Principle:
Principle #25Self-service

Solution Approach 2:

The patent implements feedback mechanisms where the automated system provides detailed verification results and confidence scores to support agents, enabling informed decision-making while reducing the time and resources agents would otherwise need to spend on manual verification, thus maintaining service quality at lower operational cost

Inventive Principle:
Principle #23Feedback

Data Source

PatentEP3476092B1Automation of image validation
Publication Date: 2020.12.16 MICROSOFT TECHNOLOGY LICENSING LLC
  • EP3476092B1 patent drawingFigure 1
  • EP3476092B1 patent drawingFigure 2
  • EP3476092B1 patent drawingFigure 3

AI summary

An automated process to determine whether an image has been modified incudes receiving an image (e.g., via a web portal), requesting an image validation service to analyze the image to determine whether the image and/or a subject depicted in the image, has been modified from its original form and, based on the analysis of the image validation service, outputting a likelihood that the image has been modified. The image validation service may analyze the image using one or more operations to determine a likelihood that the image has been modified, and provide an indication of the likelihood that the image has been modified to the web portal. The indication of the likelihood that the image has been modified may be presented on a display via the web portal, and various actions may be suggested or taken based on the likelihood that the image has been modified.