3D Luggage Threat Training Data With Optical Artifact Modeling

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Existing methods fail to provide high-quality, realistic three-dimensional training data for neural networks used in detecting alarm objects in luggage, as generic generation of such data results in unrealistic three-dimensional combination volumes due to the absence of optical artifacts, affecting the training and monitoring of operating personnel at security gates.

Innovation Solution

A method that combines object scans of exempted alarm objects with luggage scans to create a realistic three-dimensional combination volume, using electromagnetic scanning to integrate optical artifacts, ensuring the alarm object is positioned within the luggage boundaries, and adjusting for beam hardening effects, allowing for automated generation of large quantities of training data.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Quantity of substance

If three-dimensional training data is generated generically by combining exempted objects with real luggage scans, then a large quantity of training data can be provided, but the three-dimensional combination volumes become unrealistic and unsuitable for training neural networks due to absence of optical artifacts

Engineering Contradiction:

Improvequantity of training dataVSAvoidrealism of training data

Core Design Contradiction:

Quantity of substanceVSReliability

Solution Approach 1:

The patent uses copying by creating virtual three-dimensional training data through digital combination of exempted object scans with real luggage scans. Instead of physically placing objects in luggage, the system digitally superimposes object scans onto luggage scan data, generating realistic training scenarios without physical manipulation. This copying approach enables large-scale generation of training data while maintaining realism through proper integration of optical artifacts.

Inventive Principle:

Principle #26Copying

Solution Approach 2:

The patent introduces an intermediary processing system that acts as a mediator between raw scan data and final training data. This intermediary system performs several functions: it combines exempted object scans with luggage scans, applies appropriate optical artifacts and beam hardening effects, and ensures realistic positioning within luggage boundaries. This intermediary processing layer transforms generic combinations into realistic training scenarios.

Inventive Principle:

Principle #24Intermediary (Mediator)

2Ease of operation

If manual training data is generated for monitoring operating personnel, then virtual alarm situations can be created, but the realism is insufficient for effective monitoring at security gates

Engineering Contradiction:

Improvemonitoring capabilityVSAvoidrealism of virtual alarm situations

Core Design Contradiction:

Ease of operationVSReliability

Solution Approach 1:

The patent applies local quality by selectively adding optical artifacts and beam hardening effects only in the regions where exempted objects are positioned within luggage scans. Instead of uniformly processing entire scans, the system locally modifies specific areas to create realistic optical characteristics. This localized processing ensures that virtual alarm situations exhibit the same local optical properties as real threats, enhancing realism for monitoring purposes.

Inventive Principle:

Principle #3Local quality

3Productivity

If generic three-dimensional combination volumes are created, then training data can be generated quickly, but optical artifacts such as beam hardening effects are not provided, reducing training quality

Engineering Contradiction:

Improvegeneration speedVSAvoidquality of training data

Core Design Contradiction:

ProductivityVSManufacturing precision

Solution Approach 1:

The patent applies preliminary action by pre-calculating and storing optical artifact characteristics and beam hardening effect parameters for different object-material combinations. Before generating actual training data, the system prepares lookup tables and parameter sets that can be quickly applied during training data generation. This preliminary preparation enables rapid generation of high-quality training data with realistic optical artifacts without performing complex calculations during the actual combination process.

Inventive Principle:

Principle #10Preliminary action

Applied Scientific Principles

This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.

Function Achieved in This Case

Enables the creation of realistic three-dimensional training data for neural networks and monitoring personnel, enhancing the training efficiency and realism of detecting alarm objects in luggage, while maintaining cost-effectiveness.

Implementation Method 1

providing an object scan of an exempted alarm object, providing a luggage scan of an item of luggage, combining the luggage scan and the object scan into a combination scan

Methodology Applied

Scientific EffectElectromagnetic radiation transmission: Absorption (EM radiation)

Implementation Method 2

adjusting for beam hardening effects

Methodology Applied

Scientific EffectBeam hardening: Absorption (EM radiation)

Data Source

PatentUS20250329103A1Method for generating three-dimensional training data for a detection device for detecting threat objects in luggage items

Publication Date: 2025.10.23 SMITHS DETECTION GERMANY GMBH

AI summary

The disclosure relates to a method for generating three-dimensional training data for a detection device for detecting alarm objects (AO) in items of luggage (L). The method includes providing an object scan (OS) of an exempted alarm object (AO), providing a luggage scan (LS) of an item of luggage (L), combining the luggage scan (LS) and the object scan (OS) into a combination scan (CS), and generating a three-dimensional combination volume (CV) from the combination scan (CS).