Precious metal surface identification system
A high-resolution optical scanning system with deep neural networks and logistic regression improves precious metal identification accuracy and authenticity verification, addressing issues of variability and counterfeiting.
RU2865654C1Active Publication Date: 2026-07-07AVTONOMNAYA NEKOMMERCHESKAYA OBRAZOVATELNAYA ORGANIZATSIYA VYSSHEGO OBRAZOVANIYA SKOLKOVSKIJ INST NAUKI I TEKHNOLOGIJ
View PDF 4 Cites 0 Cited by
Patent Information
- Authority / Receiving Office
- RU · RU
- Patent Type
- Patents
- Current Assignee / Owner
- AVTONOMNAYA NEKOMMERCHESKAYA OBRAZOVATELNAYA ORGANIZATSIYA VYSSHEGO OBRAZOVANIYA SKOLKOVSKIJ INST NAUKI I TEKHNOLOGIJ
- Filing Date
- 2025-09-08
- Publication Date
- 2026-07-07
AI Technical Summary
Technical Problem
Existing systems for identifying precious metal surfaces face issues with low identification accuracy due to changes in image acquisition conditions, inability to determine surface uniqueness, and lack of robustness against counterfeiting, particularly in luxury goods.
Method used
A system using high-resolution optical scanning, deep neural networks for keypoint detection and matching, and logistic regression-based classification to generate and compare digital fingerprints of precious metal surfaces, incorporating anti-glare filters and glare detection.
Benefits of technology
Enhances identification reliability and accuracy, automates the process, and provides non-invasive, counterfeit-resistant authentication of precious metals and their fragments.
✦ Generated by Eureka AI based on patent content.
Smart Images
Figure 00000001_ABST
Abstract
FIELD: computer vision.SUBSTANCE: invention relates to systems for identifying the surface of precious metals using computer vision and machine learning methods. The system for identifying the surface of noble metals comprises: an optical module in the form of a visible spectrum camera capable of scanning the surface of noble metals; a positioning mechanism associated with the optical module and providing for its movement during scanning of the surface of the noble metal; a data processing server connected to the optical module and including: a) forming a digital fingerprint of the surface by detecting key points in the image and calculating descriptors for them using a first neural network model, wherein the digital fingerprint is a structured set of data corresponding to the microrelief of the surface; b) comparing the formed digital fingerprint with a reference digital fingerprint using a second neural network model to form a set of coincidence features based on the geometric and statistical characteristics of the distribution of matched key points, which include: the ratio of the moments of inertia of the distribution of key points; the density of matching points per unit area; the displacement of the centres of mass of the key points; the probability of identity of the key points (confidence score) obtained from the comparison model; c) determining the degree of coincidence of surfaces and their identification based on the calculated set of features using a classifier based on a linear machine learning model, including logistic regression.EFFECT: increase in the accuracy of identification of the surface of precious metals.5 cl, 2 dwg
Need to check novelty before this filing date? Find Prior Art