Facial Recognition System (FRS) is such an advanced technology. FRS application automatically identifies a person from a digital image. It is the ability to recognise a person by his facial characteristics. The technology is based on Eigenface (standardised faced ingredients) algorithm, which maps the characteristics of a person's face into a multi-dimensional face shape. An emerging trend in FRS is three-dimensional face recognition.

In the mid-1960s researches started their mission to recognise human face by developing special purpose software. Since then facial recognition software have survived several technical drawbacks to reach its present standard.

At its early stage, facial recognition software was capable of working with 2D images only. In order to generate an accurate result, the image of a person has to resemble all facial orientation of the stored image in a database with slight variance of light.

This scheme in many circumstances produced unwarranted results. As a result, 3D facial recognition has come into reality. Currently, nearly all facial recognition software utilise 3D model, which promises more accuracy to detect any image.

3D facial recognition system depends on some parameters to find out the desired person at any real time condition. Form the captured image the software looks at where rigid tissues and bones are most apparent: the curves of the eye socket, nose and chin.

These areas remain unchanged over time.

As 3D facial recognition requires mathematical measurements, different lighting condition does not hamper its processing to a precise output. The software depends on cumulative steps such as detection, alignment, measurement, representation, matching, verification and surface texture analysis to verify the identity of an individual.

Detection tactics include scanning image or direct photograph from the camera. This is basically the input part of the entire system. After face detection, the software determines the position, size and pose of the head. The next stage is measuring the curves of the face on a sub-millimetre scale to create a template. The template then converts into a unique code. This coding gives each template a set of numbers to represent the features on a subject's face.

If both the captured and stored images are 3D then matching will go on without further modification. But problem arises in case of 2D images. At present database contains 2D images whereas cameras capture real-time 3D images. For this, 3D images need some modifications so that they can be compared with flat and stable 2D images. For example, when a 3D image is taken, the system identifies different points in it and an algorithm will be applied to convert it to a 2D image. The software will then compare the image with the 2D images stored in the database to find an exact match.

A Minnesota based company, Identix, developed a product named Facelt@Argus that uses skin biometrics to identify any individual. This software is able to capture any image from a messy situation and compare it to a database of stored images. This matching technology combines facial geometry and skin texture for maximum accuracy.

In this process, an image of a patch of skin is captured and then segmented into smaller blocks. It is called skin print. Then an algorithm is applied to transform the patch into a mathematical, measurable space. The system then distinguishes any lines, pores and actual skin texture. This process helps doctors to identify differences between identical twins, which is not yet possible using facial recognition software. Identix claims that this combination of facial recognition and surface texture analysis increases accuracy by 20 to 25 percent.

Facelt currently uses three different templates such as vector, local feature analysis and surface texture analysis to confirm or identify the subject. This combination gives huge advantage to Facelt over other systems in case of some facial changes.

Applications of facial recognition systems are very diversified. Currently ATM employs it to determine a customer's face. When a client attempts to withdraw cash, the inbuilt camera in ATM machine captures their image. The Facelt software then generates a face print of the photograph to protect customers against identity theft and fraudulent transactions. In this process a customer does not require submitting their personal details as they do in traditional process.

Recently, the US government initiated a scheme called US-VISIT (United States Visitor and Immigrant Status Indicator Technology) to verify that travellers are who they say they are and do not pose a threat to the United States. According to this policy, certain non-US citizens who wish to enter the United States have their two index fingers digitally scanned and a digital photograph taken at the US port of entry. Immigration officials have the ability to instantly check the criminal background of the person seeking entry. Banks are also using this technology to avoid unforeseen consequences.

So far the performance of facial recognition system is quite impressive although it has got some drawbacks. But continuous research on this technology will enhance its accuracy and capacity in the near future.

Reference: howstuffworks.com