Fast, Reliable Recognition

For fast and reliable authentication, Sensible Vision has developed Continuous Adaptive Sensing™ (CAS) technology. The CAS algorithms were developed by Sensible Vision and differ dramatically from existing Facial Recognition solutions. Like other facial recognition technology, CAS makes a numeric comparison between the current biometric template values most recently acquired to a possible match for each user in the FastAccess biometric database. There the similarity ends.

Unless desired for auditing, no actual video images are permanently stored on the PC or network. Several specific patent pending technologies have been developed to enable CAS’s fast and highly accurate recognition. CAS creates each numerical template by measuring 150-1000 features on a user’s face. Particular emphasis is placed on the eyes and nose, providing greater tolerance to changes in hair style, head position and facial expression. With so many points of comparison, CAS has the detail to minimize the likelihood of incorrect recognition.

Just like humans who are better able to recognize another person the more times they meet them, CAS becomes more “familiar” each time a user accesses the device with FastAccess. CAS constantly monitors the actual time it takes to recognize each user. If the recognition time is greater than just a few seconds, CAS automatically updates the biometric database, helping to ensure faster access the next time. This adaptive learning allows FastAccess to improve its recognition database for illumination, location changes and can accommodate many day to day physically changes such as beard growth and hair style.

Sometimes the changes to a person’s face may be dramatic. For example, a user may wear contacts one day and dark sunglasses glasses the next day. If a positive match cannot be made, FastAccess prompts the user for their password or other credential. If the proper credentials are entered, the database is updated; improving recognition the next time the user access the device. Soon FastAccess will be able to recognize the user with glasses and without, and in many different lighting conditions and locations.

CAS is regularly tested for accuracy

Sensible Vision has spent a great deal of time ensuring that its CAS recognition algorithm provides fast and reliable performance. Fundamental to its design, CAS makes the most exacting matches. This minimizes false positives inherent in other biometrics that would grant the wrong person access.

In addition to extensive field use and testing, Sensible Vision has tested the CAS algorithm against the image set in a database generated by a United States Government Standards and Testing Institute. These images were created as a standard to measure the performance of existing facial recognition solutions. A given image within this database is compared to the rest of the images in the database. This test scenario is identical to the normal operation of the FastAccess software. Approximately 1,000,000 unique comparisons are made during this test. In this tests, the CAS recognition algorithm has an error rate of less than 7E-06. This is hundreds of times fewer false recognitions than the published results from the government’s public testing and far better than real life experience using passwords.

While already highly accurate, no security technology is infallible. Sensible Vision continues to enhance its technology to maintain industry leading performance of speed and accuracy.

CAS is different from traditional Facial Recognition technology.

CAS takes a dramatically different approach to Face recognition. Because of their intended applications, most Face Recognition algorithms strive for a very high recognition rate by normalizing the incoming images. This leads to a higher rate of false acceptance (recognizing the wrong person). In contrast, CAS has been developed from the ground up for continuous computer access control, where such false acceptances are unacceptable. As a result, CAS minimizes image normalization and has much lower False Acceptance rate.

A lower False Acceptance rate typically means an unacceptably high False Rejection Rate (were an authorized person is not recognized). CAS offsets the higher False Reject rate by automatically updating of the biometric database if an authorized user is not promptly recognized. Over a brief time this virtually eliminates the problem of False Rejection.

Protection against photograph access.

A photo of an authorized user is not likely to be recognized. The biometric database created by CAS was specifically designed to be sensitive to the particular lighting, angle of view, head position and numerous environmental factors. A photograph must match all these conditions and be precisely sized and placed relative to the camera. To further enhance security, FastAccess can periodically challenge for another security factor such as a Smart Card, pass phrase or other biometric technology.

CAS is able to discriminate between most identical twins.

While identical twins may look identical, the way they position and present themselves in front of the PC and web camera will likely have differences that CAS is sensitive to. In a highly secure environment where it is likely that identical twins will frequently access the same PC, Sensible Vision suggests an additional security factor such as a periodically challenging for a pass phrase, Smartcard, or other factor.