The algorithm uses GIB (Geodesic Illumination Basis) descriptors as registered data, which is calculated from 3D facial data, enabling optimal description of various pose and illumination changes.This results in superior identification accuracy essential to real world application of facial recognition technology.
The features of this technology, in comparison to NEC conventional models, are as follows:
|(1)||NEC has developed a matching method that identifies and highlights individual facial characteristics of 2D images and compares them with 3D and GIB facial data achieving 24 times better accuracy.|
|(2)||Processing time of data comparisons between 2D and 3D facial images is trebled due to NEC’s own IOF (Ideal Optical Flow) method.|
|(3)||The new algorithm realizes the world’s most accurate personal identification matching rate of 96.5%, even under very severe environmental conditions that cause changes in illumination and pose. (Tests were conducted using a database with registered 3D facial images of 1000 people.)|
Conventional 2D facial image recognition technologies typically compare registered 2D facial image data with live 2D facial image data. This can lead to unstable matching results due to the deviation in environmental conditions (such as light and posture) when the images are captured. This has been a huge barrier to accurate personal identification to date. The new NEC 3D facial recognition algorithm, has not only solved this problem, but has also succeeded in achieving greater accuracy under severe environmental conditions.
Interest in biometrics is growing on a global scale due to increasing security concerns and unstable social conditions. NEC will continue further development of this new 3D facial recognition technology to realize more flexible and even more superior real world applications.
This achievement will be demonstrated at CEBIT in Hanover, Germany from March 18 to 24, 2004, and at the General Conference of The Institute of Electronics, Information, and Communication Engineers, held at Tokyo Institute of Technology, Japan from March 22 to 25, 2004.
- Sunglasses, masks won’t fool this facial recognition — and it’s cheaper to run
- Facial recognition has a race problem — here’s how Gyfcat is fixing that
- Horizon Robotics’ smart security camera uses A.I. for serious facial recognition
- LED baseball cap fools facial-recognition tech into thinking you’re someone else
- Optical vs. laser mouse