3D Trace Imaging and Machine Learning for Production Analysis
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for analyzing handwritten or printed traces, such as in forensic science and biometrics, are limited by the loss of dynamic information in 2D analysis, reliance on human expertise, and susceptibility to data quality issues, leading to unreliable and time-consuming conclusions.
Innovation Solution
A method using three-dimensional imaging and deep learning to extract and analyze trace features, including kinematic information, to enrich the analysis with spatio-temporal data and automate the process.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of information
If 2D scanning is used to capture trace images, then the analysis can be performed on digitized documents, but the dynamic information of trace production is lost
Solution Approach 1:
The patent transitions from 2D scanning to 3D imaging to capture trace information. The 3D imaging system records depth and spatial variations that are invisible in 2D images, preserving dynamic production information such as pressure variations and stroke sequences while enabling comprehensive trace analysis.
2Reliability
If manual analysis by human experts is performed, then comprehensive trace authentication is possible, but the process is time-consuming and subjective
Solution Approach 1:
The patent replaces manual expert analysis with an automated computer-based system that processes 3D trace images. The system extracts multiple features including spatial, textural, and kinematic characteristics, then applies machine learning algorithms to authenticate traces objectively and rapidly, eliminating human subjectivity and fatigue while maintaining high reliability.
Solution Approach 2:
The system performs self-analysis by automatically extracting features from 3D images and applying trained algorithms to generate authentication results without requiring continuous human intervention. The machine learning model has been pre-trained on extensive trace data, enabling it to independently evaluate new traces with high accuracy and consistency.
3Loss of information
If multiple analysis techniques are combined for complete trace analysis, then more information can be recovered, but the complexity and time required increase
Solution Approach 1:
The patent merges multiple analysis techniques into a unified 3D imaging and analysis system. It combines spatial feature extraction, textural analysis, kinematic information recovery, and machine learning classification into a single integrated workflow that processes all trace characteristics simultaneously, reducing the need for separate analytical equipment and procedures.
Solution Approach 2:
The 3D imaging system serves multiple functions: it captures spatial geometry, measures pressure variations through depth information, determines stroke sequences through kinematic analysis, and provides textural characteristics. This multi-functional approach replaces numerous specialized devices with a single versatile system that comprehensively analyzes all trace features.
4Quantity of substance
If repeated manipulations of the original trace are performed for analysis, then multiple measurements can be obtained, but the trace quality deteriorates
Solution Approach 1:
The system creates a digital 3D copy of the trace through non-contact imaging, allowing unlimited analysis of the replica without affecting the original. Multiple measurements and re-analyses can be performed on the digital copy indefinitely, preserving the original trace quality while providing abundant measurement data for comprehensive evaluation.
Data Source
Figure 1
Figure 2~4(b)
AI summary
The invention relates to a method for generating at least one information about the production of a handwritten, hand-affixed or printed trace on a surface, comprising: - extracting several features describing the trace from at least one three- dimensional image of said trace, acquired by an imagery system, and - inputting said extracted features in a trained module to output said at least one information, said module having been trained beforehand with a plurality of previously- acquired three-dimensional images of traces and corresponding information related to the production of these traces.