Biometric System Functionality

Any biometric system should satisfy the following conditions for the biometric character chosen.

Universality: The biometric character which would be chosen must be present in all the individuals, and it should perform satisfactorily for a larger group. The fingerprint biometric character may not be suitable for manual labor with cuts and bruises.

Uniqueness: The biometric features are used for the identification of an individual which should be unique. For example, facial features of identical twins can be same and thus this methodology is considered unique but for even identical twins the palm print, fingerprint differs in their textures and features.

Permanence: The biometric character should become stable over some time. It should not be subjected to significant distinctions. Facial featured lead to variations due to aging, there may occur differences in the fingerprints due to manual work, changes in biometric character can be seen due to illness, and disease, etc. are significant limitations to fulfill the permanence requirement.

Collectability: The process of acquirement of the characters must be user-friendly and easily collected and should have high acceptability. Face recognition is user-friendly for a more extensive database system of the commercial application when compared to fingerprint. For secured applications such as passport and immigration data acquisition, where several biometric information is collected and is stored in the database, and in these applications, acceptability may not be the criteria as it becomes mandatory.

Performance: The biometric system should achieve the desired accuracy for which it was designed. Performance of biometric system is prone to numerous errors, i.e., failure to control (FTE), false accept rate (FAR), and false reject rate (FRR). The accuracy of a biometric system is not static, but it's data dependent and influenced by several factors like biometric quality of the image, size of the database, the robustness of employed algorithm, etc.

Acceptability: The biometric character must be chosen for a specific application and should have high acceptability by the user. It should be user-friendly especially for commercial and civilian applications. Face and fingerprint are universally accepted biometric when compared to iris recognition.

Circumvention: The biometric system should be foolproof. Today, with modern gadgets and with the invention of Science, the signatures can be forged, voice can be imitated, the similarity in faces, fingerprints, etc. Hence, the spoof attack should be taken care, and proper technologies should be used to prevent these attacks.

Biometric System Performance

There are many performance evaluation metrics in the biometric authentication system to evaluate the performance. Commonly adopted performance evaluation metrics are stated below:

• Genuine Accept Rate (GAR): It is the ratio of the number of input samples appropriately classified as authentic to the total number of positive input samples. A higher value of GAR indicates better performance.
• Genuine Reject Rate (GRR): It is the ratio of the number of input samples correctly classified as an impostor to the total number of impostor input samples. A higher value of GRR indicates better performance.
• False Accept Rate (FAR): It is the proportion of impostor input samples falsely classified as positive samples. It may be noted that FAR= 1 ? GRR. A lower value of FAR indicates better performance of a system.
• False Reject Rate (FRR): It is the proportion of many actual input samples falsely classified as impostor samples. Please note that FRR = 1 - GAR. A lower value of FRR indicates better performance of a biometric authentication system.
• Equal Error Rate (ERR): When FRR becomes similar to FAR, the ratio is called the same error rate. A lower value of ERR indicates better performance.
• Failure To Capture (FTC) or Failure To Acquire (FTA): It is the ratio of the number of times. A biometric system fails to capture the biometric sample presented to it. A lower value of FTA indicates better acquisition performance.
• Failure To Enroll (FTE): It is the ratio of the number of users that cannot be successfully enrolled in a biometric system and a total number of users presented to the biometric system. A lower value of FTE indicates better population coverage.

Challenges Faced in the Performance Of Biometric System

In the case of Non-biometric systems, unlike password-based authentication systems, they do not involve any complex pattern recognition techniques and hence almost perform accurately as intended by their system Designers. On the other hand, biometric data and their representations in biometric systems depend on the acquisition method, user's interaction with the acquisition device, acquisition environment, and in some cases variation in the traits, due to various pathophysiological phenomena.

Many factors affect the performance of a biometric system, some of them are concisely described below:

a. Inconsistent Presentation: Data captured by the sensor from a biometric trait depends upon the intrinsic characteristic of a biometric attribute, the way of biometric attribute presented and the user interaction with the acquisition interface. For example, due to change in pose, an appearance based on face recognition system may not match images successfully. Since different acquisitions may represent different poses of the face. Similarly, hand geometry measurements may be based on different projections of hand on a planar surface. Different iris/retina acquisitions may also correspond to different non-frontal projections of iris/retina on to the image planes.

b. Imperfect Data Acquisition: In practical situations, the data acquisition conditions are not perfect and cause extraneous variations in the acquired biometric sample. For example, non-uniform contact results in poor quality of fingerprint acquisition. The structure of a finger would be captured only if the ridges which belong to the part of the finger image are incomplete in physical/optical contact with the image acquisition surface or the valleys do not make any contact with the image acquisition surface. However, the dryness of the skin, shallow or worn-out ridges due to aging or genetics, skin disorders, sweat, dirt, and humidity in the air all confound the situation resulting in a non-ideal contact situation. Different illuminations may cause noticeable differences in the facial appearance. Backlit illumination may render image acquisition virtually useless in many applications.

c. Accuracy: In the biometric system there are two types of matching errors: false match and false non-match.

• False Match (or False Accept): It means that the biometric system incorrectly declares a positive match between the input pattern and a non-matching pattern, which stored in the database (while in the case of identification) or the profile associated with an incorrectly claimed identity (in the case of verification). In a biometric system, it should be minimized as possible, and ideally, it should be zero.
• False Non-match (or False Reject): It means that the biometric system incorrectly declares the failure of the match between the input pattern and a matching pattern in the database (in the case of identification) or the pattern associated with the correctly claimed identity (in the case of verification). In a biometric system, it should be minimized as possible, and ideally, it should be zero.