Multimodal Biometrics Fusion using Multi-Traits

The articles on Multimodal biometrics have focused on data sources. The articles looked at multi-sensor fusion strategy, multi-algorithm strategies and multi-instance fusion. Multi-sensor fusion strategy utilises a single biometric trait captured using multiple sensors in order to extract diverse information from the image. While, multi-algorithm system uses only a single biometric capturing device (single sensor) to obtain raw data and then the raw biometric data is processed using multiple algorithms. It is clear that multi-algorithm systems are cheaper and more convenient for customers than multi-sensor as multi-algorithm systems do not require the use of additional biometric capturing devices. The customers are not required to interact with multiple capturing devices. However, multi-algorithm fusion requires the introduction of a new feature extractor and/or matcher modules which may increase the computational requirements of the system. Multi-instance fusion strategy utilises multiple instances of the same body trait such as the left and right index fingers of an individual may be used to verify an individual’s identity. Multi-instance fusion strategy overcome the limitations of both multi-sensor and multi-algorithms as Multi-instance does not introduce new sensors nor does it require the development of new feature extraction and matching algorithms.

In this article we introduce multi-traits fusion strategy. Multi-traits fusion strategy utilises multiple body trait to establish identity. For example, multi-traits biometric systems can utilize face and voice features to establish the identity of an individual. Uncorrected biometric characteristics are expected to result in better improvement in performance than correlated biometric traits. Voice and lip movement are example of correlated traits while fingerprint and iris are examples of uncorrelated biometric traits. The identification accuracy can be significantly improved by utilizing an increasing number of biometric characteristics. The number of biometric characteristics used in a specific application will be restricted by practical considerations such as the cost of deployment, enrollment time, throughput time, expected error rate, user habituation issues. Generally, the main limitations of multi-traits are cost and the development of appropriate user interfaces.

Other fields – Information fusion continue to be used in a diversity of scientific fields such as Land Mine detection. Several types of sensor technologies are being used to detect buried land mines. These include electromagnetic induction (EMI), ground penetrating radar (GPR), infra-red imaging (IR), quadrupole resonance (QR), chemical detectors and sensors of acoustically induced surface vibrations. In many cases, the data presented by these multiple sensors are concurrently used to improve the accuracy of land mine detection algorithms.

Our researchers within BRL at Namibia Biometric Systems will continue to examine the levels of fusion that are possible in a multimodal biometric system in the next article. In addition, our researchers at BRL would like to thank all the readers of the articles from Namibia, Africa and world wide.

More information on the implementation of biometrics based solutions can be requested from info@namibiabiometricsystems.com.

Multimodal Biometrics Fusion using Multi-Instance

The articles on Multimodal biometrics have focused on data sources. The articles looked at multi-sensor fusion strategy, multi-algorithm strategies and multi-instance fusion. Multi-sensor fusion strategy utilises a single biometric trait captured using multiple sensors in order to extract diverse information from the image. While, multi-algorithm system uses only a single biometric capturing device (single sensor) to obtain raw data and then the raw biometric data is processed using multiple algorithms. It is clear that multi-algorithm systems are cheaper and more convenient for customers than multi-sensor as multi-algorithm systems do not require the use of additional biometric capturing devices. The customers are not required to interact with multiple capturing devices. However, multi-algorithm fusion requires the introduction of a new feature extractor and/or matcher modules which may increase the computational requirements of the system. Multi-instance fusion strategy utilises multiple instances of the same body trait such as the left and right index fingers of an individual may be used to verify an individual’s identity. Multi-instance fusion strategy overcome the limitations of both multi-sensor and multi-algorithms as Multi-instance does not introduce new sensors nor does it require the development of new feature extraction and matching algorithms.

In this article we introduce multi-modal fusion strategy. Multi-modal fusion strategy utilises multiple body trait to establish identity. For example, multimodal biometric systems can utilize face and voice features to establish the identity of an individual. Uncorrected biometric characteristics are expected to result in better improvement in performance than correlated biometric traits. Voice and lip movement are example of correlated traits while fingerprint and iris are examples of uncorrelated biometric traits. The identification accuracy can be significantly improved by utilizing an increasing number of biometric characteristics. The number of biometric characteristics used in a specific application will be restricted by practical considerations such as the cost of deployment, enrollment time, throughput time, expected error rate, user habituation issues. Generally, the main limitations of multi-modal are cost and the development of appropriate user interfaces.

Multi-instance overcomes the limitations of both multi-sensor and multi-algorithms. Multi-instance fusion strategy has the limitations that it may be slow and may result in poor customer satisfaction. Researchers at BRL highlights that the limitations of multi-instances can be simply overcome by a new sensor arrangement in order to facilitate the simultaneous capture of the various units/instances. Multi-instance fusion strategy is advantageous for individuals whose biometric characteristics cannot be reliably captured due to inherent problems. For example, an Automated Fingerprint Identification Systems (AFIS), may benefit from sensors that are able to rapidly acquire impressions of all ten fingers. NBS researchers understand that a single finger may not be a sufficient discrirninator for a person having dry skin or bad fingerprints. Therefore, the integration of evidence across multiple fingers may serve as a good discriminator. Similarly, an iris system may not be able to image significant portions of a person’s iris due to drooping eyelids.

Other fields – Information fusion continue to be used in a diversity of scientific fields such as Object detection. Many applications attempt to detect and establish the trajectories of objects based on the evidence supplied by multiple image modalities. The fusion of visible and non-visible information pertaining to different wavelengths in the electromagnetic spectrum such as radar and infrared images, thermal and visible spectrum images may assist in estimating the location and kinematic features of objects such a squad of soldiers in a night-time battlefield. These applications rely on image fusion methodologies to combine multiple modalities.

Our researchers within BRL at Namibia Biometric Systems will continue to examine the levels of fusion that are possible in a multimodal biometric system in the next article. In addition, our researchers at BRL would like to thank all the readers of the articles from Namibia, Africa and world wide.

More information on the implementation of biometrics based solutions can be requested from info@namibiabiometricsystems.com.

Information Fusion For Multimodal Biometrics

Multimodal biometric systems overcome limitations such as:

Noise in captured data – A voice sample can be altered by cold while a fingerprint image can be altered by a scar. Noise data can also be a result of defective or improper maintenance of the sensors. When the captured biometric data from a single trait is corrupted with noise, the availability of other (less noisy) biometric characteristics may aid in the reliable determination of identity.

Intra-class variation – This is typically due to an individual who incorrectly interacting with the sensor or due to changes in the biometric characteristics of a person over a period of time. Therefore, results in a variation of the biometric characteristic for the same individual. Intra-class variation can be handled by storing multiple templates for every user.

Inter-class similarities – This refers to the overlap of discriminating features corresponding to multiple individuals. Increasing the number of individuals enrolled in a biometric system, is likely to increase the inter-class similarities and false match rate of an identification system. Multimodal biometric increases the number of discriminatory features for every individual.

Non-universality – The biometric system may not be able to capture meaningful biometric data from a subset of users. A fingerprint system may extract incorrect minutia features from the fingerprint of certain individuals due to poor quality of the ridges. Multimodal can substantially improve the accuracy of a biometric system depending upon the information being combined and the fusion methodology.

Spoof attacks – This involves the deliberate manipulation of one’s biometric characteristic in order to avoid recognition or the creation of a fake biometric characteristic in order to take on the identity of another person. It becomes increasingly difficult for an impostor to spoof multiple biometric characteristics of a legitimately enrolled individual.

The inherent limitations of a single biometric system are alleviated by fusing the information presented by multiple sources. Fusion of biometric information results in increasing the number of discriminating features used to represent individuals. The magnitude of discriminatory information available in a biometric characteristic is restricted. Therefore, it increases the number of individuals who can be enrolled in the biometric system.

Humans recognize each other using multiple behavioural and physical biometric characteristics. Therefore, the recognition process may be viewed as the reconciliation of evidence from multiple modalities. Although humans can sometimes use a single modality to identify each other, it is not always used to reliable perform recognition.

Information fusion has also been used in a diversity of scientific fields such as weather forecasting. Weather forecasting system relies on evidence provided by diverse sources of information such as geostationary meteorological satellites, weather balloons/planes, ground stations, etc.

In this article we consider multi-sensor fusion strategy. Multi-sensor system uses a single biometric characteristic captured by multiple sensors in order to extract diverse information. For example, a system may capture the two dimensional texture content of a person’s face using a 2D camera and a three dimensional surface shape of the face using a 3D camera. The use of multiple sensors, in some instances, can result in the acquisition of complementary information that can enhance the recognition ability of the system. The performance of a 2D face matching system can be improved by utilizing the shape information presented by 3D range images.

Our researchers within Biometric Research Laboratory, BRL, at Namibia Biometric Systems will continue to examine the levels of fusion that are possible in a multimodal biometric system in the next article.

More information on the implementation of biometrics based solutions can be requested from info@namibiabiometricsystems.com.

Levels of Fusion in Multimodal Biometric Systems

The previous articles on Multimodal biometrics had focused on data sources such as multi-sensor, multi-algorithm, multi-instance and multi-traits. This article will focus on the levels of fusion. The amount of information available to a biometric system drastically decreases as one proceeds from the sensor module to the decision module (matching module). In a multibiometric system, fusion can be accomplished by utilizing the information available in the sensor module, feature module, score module and decision module. The levels of fusion can be broadly classified as fusion prior to matching and fusion after matching.

Fusion prior to matching: This is the fusion of data prior to matching. The integration of information from multiple biometric sources can take place either at the sensor level or at the feature level. The raw data from the various sensors can be integrated or combined in sensor level fusion. Sensor level fusion can only be employed if the multiple sources represent samples of the same biometric characteristics obtained either using a single sensor or different compatible sensors. For example, 2D face images of an individual obtained from several cameras can be combined to form a 3D model of the face. While Feature level fusion refers to the integration of different feature sets extracted from multiple biometric sources. Biometric features used in the feature level fusion can be divided into two set, homogeneous and non-homogeneous. When the feature sets are homogeneous (e.g., multiple measurements of the same biometric characteristic such as a person’s hand geometry), a single resultant feature vector can be calculated as a weighted average of the individual feature vectors. While if the feature sets are non-homogeneous (e.g., features of different biometric modalities such as face and hand geometry), the features can be concatenated into a single feature vector. Feature selection schemes are employed to reduce the dimensionality of the ensuing feature vector. Concatenation is not possible when the feature sets are incompatible.

Fusion after matching: This is the fusion of data post matching. The integration of information after the matcher stage can be divided into four categories: dynamic classifier selection, fusion at the decision level, fusion at the rank level and fusion at the match score level. A dynamic classifier selection scheme chooses the results of that biometric source which is most likely to give the correct decision for the specific input pattern. This is also known as the winner-take-all approach within the biometrics world. Integration of information at the decision level can take place when each biometric system independently makes a decision about the identity of the user (in an identification system) or determines if the claimed identity is true or not (in a verification system). Since most commercial biometric systems provide access to only the final decision output by the system, fusion at the decision level is often the only viable option. When the output of each biometric system is a subset of possible matches (i.e., identities) sorted in decreasing order of confidence, the fusion can be done at the rank level. This is relevant in an identification system where a rank may be assigned to the top matching identities. When each biometric system outputs a match score indicating the proximity of the input data to a template, integration can be done at the match score level.

More information on the implementation of biometrics based solutions can be requested from info@namibiabiometricsystems.com.

Multimodal Biometrics Fusion using Multi-Instance

The previous two articles focused on multi-sensor fusion strategy and multi-algorithm strategies. Multi-sensor fusion strategy utilises a single biometric trait captured using multiple sensors in order to extract diverse information from the image. On the other hand, multi-algorithm systems circumvent the limitations of multi-sensor as multi-algorithm systems do not require the use of additional biometric capturing devices and the associated costs. In addition, the users are not required to interact with multiple capturing devices and thus enhance user convenience. However, NBS researcher at the Biometric Research Laboratory, BRL, are keen to highlight multi-algorithm fusion requires the introduction of a new feature extractor and/or matcher modules which may increase the computational requirements of the system.

In this article we introduce multi-instance fusion strategy. Multi-instance fusion strategy utilises multiple instances of the same body trait. This can be illustrated using multiple examples, the left and right index fingers of an individual may be used to verify an individual’s identity. However, multi-instance fusion strategy does not impose the introduction of new sensors nor does it entail the development of new feature extraction and matching algorithms. Similarly, the left and right iris of an individual does not require the introduction of new sensors or multiple algorithms which introduces associated costs and computational complexity. Multi-instance overcomes the limitations of both multi-sensor and multi-algorithms. Multi-instance fusion strategy has the limitations that it may be slow and may result in poor customer satisfaction. Researchers at BRL highlights that the limitations of multi-instances can be simply overcome by a new sensor arrangement in order to facilitate the simultaneous capture of the various units/instances. Multi-instance fusion strategy is advantageous for individuals whose biometric characteristics cannot be reliably captured due to inherent problems. For example, an Automated Fingerprint Identification Systems (AFIS), may benefit from sensors that are able to rapidly acquire impressions of all ten fingers. NBS researchers understand that a single finger may not be a sufficient discrirninator for a person having dry skin or bad fingerprints. Therefore, the integration of evidence across multiple fingers may serve as a good discriminator. Similarly, an iris system may not be able to image significant portions of a person’s iris due to drooping eyelids.

Other fields – Information fusion continue to be used in a diversity of scientific fields such as Object detection. Many applications attempt to detect and establish the trajectories of objects based on the evidence supplied by multiple image modalities. The fusion of visible and non-visible information pertaining to different wavelengths in the electromagnetic spectrum such as radar and infrared images, thermal and visible spectrum images may assist in estimating the location and kinematic features of objects such a squad of soldiers in a night-time battlefield. These applications rely on image fusion methodologies to combine multiple modalities.

Our researchers within BRL at Namibia Biometric Systems will continue to examine the levels of fusion that are possible in a multimodal biometric system in the next article. In addition, our researchers at BRL would like to thank all the readers of the articles from Namibia, Africa and world wide.

More information on the implementation of biometrics based solutions can be requested from info@namibiabiometricsystems.com.

Multimodal Biometrics Fusion using Multi-Algorithms

The previous article focused on multi-sensor fusion strategy where a single biometric trait is captured using multiple sensors in order to extract diverse information from the image. A multi-sensor fusion strategy can be implemented as follows using face and fingerprint traits:

Face Biometric System – a system may capture the two dimensional texture content of a person’s face using a 2D camera and a three dimensional surface shape of the face using a 3D camera. The use of multiple sensors, in some instances, can result in the acquisition of complementary information that can enhance the recognition ability of the system. The performance of a 2D face matching system can be improved by utilizing the shape information presented by 3D range images.

Fingerprint Biometric System – a system may capture an individual’s fingerprint images using an optical fingerprint sensor which involves capturing a digital image of the print using visible light ( a specialized digital camera) and capacitive fingerprint sensor which use principles associated with capacitance in order to form fingerprint images. The two sensors provide complementary information and therefore results in enhanced matching accuracy.

Although a multi-sensor strategy has the essential benefits of enhanced accuracy if implemented corrected, the introduction of additional biometric capturing equipments such as a 3D camera to measure the facial surface variation and optical sensor for fingerprint increases the cost of the multimodal biometric system.

Unlike multi-sensor systems highlighted above and in the previous article, in this article we consider a multi-algorithm fusion strategy. Multi-algorithm system uses only a single biometric capturing device (single sensor) to obtain raw data and then the raw biometric data is processed using multiple algorithms. Multi-algorithm fusion strategy can be applied to an Automated Fingerprint Identification System (AFIS) as follows:

AFIS – the same fingerprint image captured using an optical sensor can independently be processed by texture-based fingerprint algorithm and a minutiae-based fingerprint algorithm in order to extract diverse feature sets that can improve the performance of the system.

Multi-algorithm systems do not require the use of additional biometric capturing devices and therefore overcomes the limitations of the Multi-sensor system such as costs. Multi-algorithm strategy can be cost-effective. Furthermore, the user is not required to interact with multiple capturing devices and therefore multi-sensor fusion strategy can enhance user convenience. However, multi-sensor fusion requires the introduction of new feature extractor and/or matcher modules which may increase the computational requirements of the system.

Other fields – Information fusion has also been used in a diversity of scientific fields such as Robotics for navigation. Generally a robot is fitted with a variety of sound, light, image and sensors that allow it to record its environment. In order to determine a suitable action such as move right or tilt camera at a certain angle, the data acquired using these multiple sensors are processed simultaneously.

 

Our researchers within Biometric Research Laboratory, BRL, at Namibia Biometric Systems will continue to examine the levels of fusion that are possible in a multimodal biometric system in the next article.

More information on the implementation of biometrics based solutions can be requested from info@namibiabiometricsystems.com.