Counterfeit coins can be detected more easily thanks to a novel approach

Steel cash could also be simply in regards to the oldest medium of change nonetheless in use at present, however guaranteeing their price requires a few of the most state-of-the-art know-how out there. Counterfeit cash stay a menace to world currencies, with malicious actors flooding markets with fakes. European police broke up a Spain-based prison ring in late April, demonstrating the difficulty’s ongoing urgency.

Nonetheless, no counterfeit is totally detection-proof, regardless of how real it seems. There are at all times some tell-tale indicators of forgery, even when they aren’t instantly noticeable to the bare eye.

In a brand new paper within the journal Skilled Methods With Purposes, a gaggle of Concordia researchers on the Centre for Sample Recognition and Machine Intelligence (CENPARMI) current a novel framework that makes use of image-mining strategies and machine studying algorithms to establish flaws in counterfeit cash.

“Using image technology, we scanned both genuine and counterfeit coins so we can look for anomalies that may be either two- or three-dimensional, such as letters or the face of the person on the coin,” says Ching Suen, a professor within the Division of Laptop Science and Software program Engineering and the paper’s supervising creator.

“This framework is not only about safeguarding our economy and resources,” provides the paper’s lead creator, CENPARMI postdoctoral fellow Maryam Sharifi Rad. “It is also about pushing the boundaries of technology and improving security.”

CENPARMI analysis affiliate Saeed Khazaee additionally contributed to this research.

The tell-tale blob

The researchers’ framework is constructed round fuzzy affiliation guidelines mining. This method makes use of synthetic intelligence to seek out patterns which are related however “fuzzy,” i.e., not clear sufficient to be precise copies. Nonetheless, the framework will ultimately arrive at a sure vary of outcomes the place constructive matches be confidently recognized.

The strategy begins by utilizing state-of-the-art scanners to look at cash suspected of being counterfeit. The cash are offered by legislation enforcement companies.

The scanned photographs are then segmented to areas of curiosity, which encompass collections of localized coherent areas known as “blobs.” These blobs are acknowledged based mostly on visible similarity and composition, which give related options the researchers can extract. Blobs are like clues that assist the researchers work out what’s going on within the scanned photographs.

Fuzzy affiliation guidelines mining is carried out utilizing these blobs to extract frequent patterns from the pictures. These patterns seize relationships among the many blobs’ attributes, equivalent to colour, texture, form and dimension. The patterns assist researchers to higher perceive the pictures and inform whether or not a coin is actual or faux.

The blobs play an important function in producing fuzzy affiliation guidelines, which signify implicit data and relationships inside a set of photographs, aiding in picture classification duties.

The researchers say they imagine their method will be utilized to detect all method of counterfeit objects past cash.

“This method can be used to detect all kinds of fake goods, which we are seeing all over the world,” says Suen. “It can also be used to detect fake labels on fruits, wines, liquor and so on. There are many places where this can be applied.”