Berbix’s virtual authentication detection powering Socure’s DocV granted patent
Forensics and data extraction developer Berbix, which was acquired by Socure just months ago, has received a patent for their “virtual authentication detection” technology that powers DocV 3.0, Socure’s biometric document verification software.
Other document verification methods may use a template library of all U.S. IDs. “But one of the challenges with that method is that fake ID manufacturers are able to put the appropriate holograms and signals on there to make it look exactly like a legitimate ID,” said Eric LeVine, one inventor of the newly patented system, in an interview with Biometric Update.
Using a dataset of legitimate and fake IDs, Berbix can build systems to find discrepancies among human and machine-readable elements of ID documents to distinguish visually convincing fakes from their legitimate counterparts.
With the patented system, “we’re able to detect whether it’s a legitimately printed barcode or legitimately printed portrait of the person’s face,” and that it “legitimately matches all the information and has the consistency that we’re looking for from more of a deep data analysis perspective,” says Levine.
The patent reveals the system is a multilayered approach that can detect multiple identification and authentication indicators in virtual images of identity documents and can determine if the document is legitimate by analyzing the indicators.
The detection system generates a machine learning model consisting of a “data set of information associated with authenticated documents” to analyze identification indicators and authentication indicators to determine if an ID or document is authentic based on virtual images, according to the patent.
The system can use identification indicators to determine whether at least one or more authentication indicators is inauthentic. It can determine if identification indicators are authentic by analyzing identification indicators and authentication indicators.
Additionally, analyze a data payload from decoding identification codes and then analyze the sequence and order of the authentication fields in the payload, determining “whether a rule-based assertion is present within an authentication field,” and “comparing a plurality of authentication fields within the data payload,” the patent says.
“All of the systems that we use for data extraction and classification have all been integrated deeply into Socure’s DocV solution,” says Levine. “And so it is powering that system overall.”