Central to the technology created by Warwick Warp is a completely new approach to image recognition. Instead of relying solely on a set of features, Warwick Warp has developed an innovative modelling technology which allows the object to be described using mathematical functions. This creates an idealised model of the object, with an understanding of the underlying characteristics and structure; features can then be captured using a distortion-free coordinate system. A major advantage of this technique is that when presented with an image that varies from the ideal, the system can cope effectively with the variation and has a higher chance of producing a successful identification. The Warwick Warp modelling technology, which is far more robust to variation than traditional methods, has been patented and research is continuing to refine, improve and develop the matching algorithms.
Creating an automatic recognition system is a challenging proposition. The traditional approach to this difficult task is to "train" the recognition program using a set of parameters that describe the pattern of interest, so that when it is subsequently presented with an image of the pattern, it can make a comparison between the image and the set of parameters. Provided that the image falls within the prescribed parameters, recognition will take place, but if there is variation from the norm, it may be rejected and the object will not recognised. Add in other factors such as poor image quality, low contrast and some types of geometric distortion or warping, and the process of recognition becomes even more difficult and unreliable.
Central to the technology created by Warwick Warp is a completely new approach to image recognition. Instead of relying solely on a set of features, Warwick Warp has developed an innovative modelling technology which allows the object to be described using mathematical functions. This creates an idealised model of the object, with an understanding of the underlying characteristics and structure; features can then be captured using a distortion-free coordinate system. A major advantage of this technique is that when presented with an image that varies from the ideal, the system can cope effectively with the variation and has a higher chance of producing a successful identification. The Warwick Warp modelling technology, which is far more robust to variation than traditional methods, has been patented and research is continuing to refine, improve and develop the matching algorithms. Warwick Warp's R&D falls into two areas of Biometric research;
Fingerprint Identification
Face Recognition
Fingerprint Identification
The first application of the Warwick Warp modelling concept is a Software Development Kit (SDK) that can be licensed and integrated into commercial automated fingerprint identification systems (AFIS). Independent benchmarking has shown that the accuracy performance of the Warwick Warp fingerprint matching system is comparable to or better than established top tier AFIS companies.
For further information about this application, please visit the the Products page.
Face Recognition
An area that could potentially benefit from the Warwick Warp modelling technology is facial recognition. Currently available systems have limited success as they struggle to deal with the large variations that can occur between a stored facial image and the image that is captured when a person presents him- or herself for identification. Such variations are caused by pose (i.e. the angle that the face is presented for viewing), variation in lighting and background (brightness, direction of the light source etc.) and variations such as expression, the amount of make-up being worn, and so on.
Warwick Warp has extended its technology to address these issues by using mathematical concepts to create an idealised model of the face. When asked to identify that face from an image acquired in less than optimum conditions, or from a different angle, or wearing a smile, a mathematical model provides a much higher chance of recognition, because the system is not simply relying on a set of given facial features.
Warwick Warp has patented technology that it has developed specifically to address the difficulties caused by lighting variation. This technology can automatically estimate both the amount and direction of the light falling on an image and remove it to produce a light neutral image to present for identification and matching.
Warwick Warp is actively engaged in research and development projects designed to deliver solutions for commercial applications. The dedication and determination of the company's highly skilled team of software engineers will, in the fullness of time, lead to new, robust technologies and intellectual property to licence for use in the biometric applications of the futureCreating an automatic recognition system is a challenging proposition. The traditional approach to this difficult task is to "train" the recognition program using a set of parameters that describe the pattern of interest, so that when it is subsequently presented with an image of the pattern, it can make a comparison between the image and the set of parameters. Provided that the image falls within the prescribed parameters, recognition will take place, but if there is variation from the norm, it may be rejected and the object will not recognised. Add in other factors such as poor image quality, low contrast and some types of geometric distortion or warping, and the process of recognition becomes even more difficult and unreliable
Fingerprint Matching and Identification
This completely new approach to fingerprint matching exceeds established accuracy benchmarks and delivers ultra-high accuracy for all types of fingerprints. The performance has been independently verified (NIST PFT and ELFT tests) as being among the most accurate available.
We specialise in developing innovative software products that enable integrators and developers to incorporate highly accurate fingerprint matching into their solutions. The fingerprint matching technology is made available as Software Development Kits (SDK) and SDK has been optimised to address the requirements of specific markets: AFIS, police (latent prints) and standard LiveScan applications (access control etc.).