Automatic Biometric Signature Verification in Real Time

Typical Use Cases

  • Satisfying legal requirements in an automated, reliable and proofable manner
  • Lowering the risk of signatur forgery
  • Fraud prevention on e.g. high value transactions

Automated Dynamic Verification of Behavioral Biometrics

  • SIGNificant captures the behavioral biometric data of ones handwritten signature (including speed, acceleration, rhythm, movements in the air and pressure) and embed the signature into an electronic document.
  • The captured biometric signature data is automatically verified dynamically against signature specimen stored encrypted in the systems database.
  • The result of the automated signature verification is provided via a signed response ensuring its authenticity. Additionally a server-side audit trail keeps track of every action allowing statistical analysis on historical data or to verify a specific verification result.
  • Documents may only get signed if the authentication based on ones handwritten signauture is passed succesfully

How It Works: Three Easy Steps

  1. Enrollment (Collect Sample Signatures) - The process is initiated by the user’s enrollment. The enrollment process requires up to six signatures, which can be collected initially or over a period of time. The signatures are collected using a signature pad or a tablet; the user simply signs a personal handwritten signature, exactly as he or she would with an ordinary pen and paper. 
  2. Signature Profile Creation and Handling - The signatures are stored in profiles. The system can handle numerous multiple profiles per user, allowing, for instance, one profile for a standard signature and another for signing with initials only. When verifying a signature, the SIGNificant Biometric Server compares the signatures to the relevant profiles. 
  3. Signature Verification and Learning - Each time a user accesses the system to verify his signature, the Biometric Server compares the current signature to the signature profiles. With each authentication, the server continues to learn and fine-tune the user’s profile. This enables the system to track gradual shifts in the handwritten signature over time.

Signatures Are The Only Biometrics That Vary Over Time

  • Every biometric system, such as an iris scan or fingerprints, may be hacked. Once hacked, the information may be used again and again because eyes and fingerprints do not change - they are static. 
  • By contrast, a signature, even if hacked, is not reusable since no-one can ever sign the same signature twice; signatures are bound to be different from one another. 
  • Also, the user can always change a signature and create a new personal profile.

Award-Winning Online Signature Verification

  • At the 2011 ICDAR conference, participants' signature verification solutions were evaluated by forensic experts using different testing sets with skilled forgeries. Their task was to determine whether a particular signature had been written by the author of the reference signatures or if it had been forged by another writer. In all experiments, twelve known reference signatures were presented to the systems.
  • ICDAR evaluated the systems using to several measurements. They generated ROC-curves to see at which point an equal error rate was reached: i.e. the point where the false acceptance rate (forged signature being accepted as genuine) equals the false rejection rate (genuine signature being rejected). At this specific point they also measured the accuracy, i.e. the percentage of correct decisions with respect to all queried signatures.
  • The clear winner was SIGNificant!

How is SIGNificant Different from Other Signature Verification Solutions?

  • xyzmo offers the world’s most complete, open and accurate real-time signature verification solution. The major advantage over other solutions on the market today is that our technology is able to compare a signature against a profile that is self-learning over time. Only this approach guarantees appropriate results for signature verification and authentication, because it is human nature never to sign twice in exactly the same way, and also to alter the signature constantly over a life-time.
  • A comparison with only one sample signature is mathematically much easier to handle, and thus some companies offer solutions that are essentially inaccurate. These “low level” solutions merely pick one random signature as a basis for the comparison. This approach only works for people who always sign in exactly the same way. Most human beings do not behave like that, and thus this approach is simply not feasible for a broader usage of that technology. Ask 10 people to sign 6 times in a row, on a blank sheet of paper, and you can see for yourself how different most of thoese signatures will be.
  • A more sophisticated but still unsatisfactory approach is to build a solution that takes several signatures – a profile – into consideration at the time of real-time comparison, but still uses a static profile that is not self-learning. This will deliver, better results than a comparison with just one signature at the start, but the comparison will get less and less accurate over time. This will happen for nearly 100% of human subjects.
  • Only the xyzmo SIGNificant approach – based on self-learning profiles – provides sufficient accuracy in the long run. The SIGNificant Biometric Server takes the ability of self-learning profiles one step further, by using a sophisticated algorithm when the signature profile is being built, that assesses whether the profile is of sufficient quality or not. When people sign on a signature pad for the first time, they change their signing behavior a little bit; later they adjust it after they get used to this new technology, or the new way of signing on a signature tablet, until it becomes the “usual” way. Thus it is business-critical to take more than two or three samples for each profile and, on top of that, to check by intelligent algorithms to see if these profiles are a robust base for later verification. It is much better to reject improper signatures from a profile at the time the profile is being create, and ask a customer to sign one additional time,than to generate a poor profile and use it for later comparison.

Trusted by Leading Companies Worldwide

  • Unicredit (Italy) 
  • Banco BMG (Brasil)
  • Poste Italiane (Italy)
  • BNL Italy (Banca Nazionale del Lavoro SpA) 
  • GE Money Bank (CZ) 
  • Tatra Banka (Slovakia) 
  • CSE Italy (Banking Service Provider) 
  • Cabel Italy (Banking Service Provider) 
  • Central Bank of Honduras (CNBS) 
  • Sabemi Group (Brasil) 
  • HP purchased more than 22.000 Licenses for a large e-signature project in United Kingdom
  • NASK (Poland
  • Maquet Carediopulmonary AG (Germany)
  • and many more ...

SIGNificant Biometric Server

Works in conjunction with all SIGNificant signature capture products and SDKs

  • Enterprise Edition
  • Professional Edition
  • Small Business Edition