Thanks for visiting Imaging and Machine Vision Europe.

You're trying to access an editorial feature that is only available to logged in, registered users of Imaging and Machine Vision Europe. Registering is completely free, so why not sign up with us?

By registering, as well as being able to browse all content on the site without further interruption, you'll also have the option to receive our magazine (multiple times a year) and our email newsletters.

Recognising imaging's fingerprints

Share this on social media:

Topic tags: 

Tom Eddershaw looks at the imaging techniques used in fingerprint recognition systems, and the challenges manufacturers face in deploying the technology

The Touch ID security function on the iPhone 6 is one of the more high-profile examples of a commercial use of fingerprint recognition, and while fingerprints and other biometric forms of identification are becoming more widely used, is biometric technology ready for large-scale deployment?

Looking at commercially available reports, the general market for biometric technology is around $11bn in 2014, noted Stephen Anderson, industry and market strategist at SPIE. He said that projections have this going to $27bn by 2019, at a CAGR of around 20 per cent.

SPIE, the optics and photonics society, is in the middle of analysing a number of market segments related to defence and security, biometrics being one of them. ‘My sense is that many biometric technologies are not quite ready for prime time as far as general consumer implementation is concerned but fingerprint recognition is coming,’ Anderson commented.

So where does imaging come into all of this? The term biometrics encompasses a number of different technologies, but all use human characteristics as a security measure. Imaging is used in a number of these, including fingerprint, vein, iris, and facial and gait recognition. Applications can be found across the security market, from government level to personal security, but progress is being slowed by implementation cost and a lack of confidence in the technology, either its accuracy or the privacy issues associated with recording and storing personal information.

In order to further uptake, Emanuela Marasco, a post-doctoral associate researcher at the University of North Carolina, Charlotte’s department of computer science, believes improving public confidence in biometric technology is crucial. She explained that when considering the deployment of biometric technology, it is very much related to the acceptance degree of the users. ‘It depends on how willing the user is to give you biometric data; it’s a social concern,’ she said.

Marasco will present research at SPIE DSS, the defence, security and sensing show to be held in Baltimore, USA from 20-24 April. She explained one of the reasons for the lack of trust in the technology: ‘You can acquire fingerprint data for one purpose, but it could be used for something else. If a fingerprint biometric is stored it could be communicated to another device and then used for a purpose that might be unknown to the subject. This is obviously an invasion of privacy, because people don’t know whether their data is being used in unauthorised applications.’

Fingerprint biometric data is created from an image by logging key minutiae, details within the fingerprint used for identification, such as ridges ending, beginning or forking. Marasco explained that these points are characterised by location in terms of x and y within the image, as well as orientation. ‘This is what we want to protect. If you consider the template of a fingerprint, and you have say 100 points characterised by these minutia points, you want to protect that.’ This can be done before feature extraction, or once there is an image template.

In her talk at DSS, Marasco will explain that improving the privacy and protection of the subjects of biometric imaging will assist in improving public confidence in biometric technology in general. There are a number of different ways to go about this at different levels of the process, but the fundamental point is to remove the personal information from a fingerprint.

She continued: ‘We work straight from the original image and remove all of the sensitive information. The image looks like a typical fingerprint but it doesn’t have the personal information. After feature extraction, you are encrypting a protected template with the metric being taken within the encrypted domain. By taking the metric within the encrypted domain, the original template – and therefore the original image – can be protected.’

Another method is to de-identify the image as soon as it’s taken. The process is to remove the information about the identity of the subject of an image. Marasco said that there is already work being done where the images of fingerprints are mixed; once an image is taken, it is blended with a similar fingerprint from a database and stored, meaning the original image isn’t stored.

Characteristics such as gender or age are predictable from biometric data. So called soft biometrics are a concern for privacy, because even if all of the information about a subject has been removed, the age or gender can still be discovered and reduce the list of possible people it could be. ‘It’s still very important information to hide – de-identification is done to remove soft biometrics from fingerprints, in particular gender.’

Marasco highlighted that while certain information should be removed in order to protect the subject, the remaining print still needs to be useable and to provide results.

‘We need to consider the resolution of the images. All of these algorithms are related to a connected system, and I can say 500dpi as a typical resolution,’ she said. Other products offer different resolutions, so in order to cover most systems Marasco and her team ensure the algorithms work up to 500dpi. The higher end systems offer up to 1,000dpi but Marasco said these are very expensive. The group have done some research into whether the size of the image has any impact on their findings.

As technology improves, privacy will improve. Marasco added that it also increases the acceptance because people can trust the machines to work accurately. Now though, according to Marasco, there is more funding for privacy and it’s now one of the key areas of development in the US.

There is another issue with biometrics compared to conventional security. Marasco pointed out that a biometric is a permanent piece of information and it can’t be changed, unlike a password. Once recorded, she said, ‘somebody will have, forever, my identity.

But if an encrypted version of the biometric is created, this can be adjusted or the user could create multiple different fingerprint signatures. And this will be the research of the future of biometrics.’

Eye in the sky

Aeraccess, a French start-up focusing on unmanned aerial vehicle (UAV) R&D, have developing an affordable unmanned surveillance system, making the most of military technology combined with Sony’s FCB-EV7500 camera. The technology can be used to get a bird’s eye view of a car accident on a highway, fire in a forest or in a building; support policemen during outdoor activities, crowd control, and search and rescue; or remote area patrol, or border security.

Nick Waple, product marketing specialist for Sony’s Image Sensing Solutions Europe division, commented: ‘Aeraccess is taking a PTZ camera and making it fly, which in itself is impressive. The fact that they are aiming at the oil and gas, and the fire and rescue sectors signifies that this is a company with ambition and capability. It also illustrates the flexibility and quality of Sony FCB camera modules. Aeraccess’s product needs to deliver great image quality, reliably, in tough conditions. Sony HD FCB camera modules excel at exactly this.’

Waple added: ‘The UAV market is booming in various segments: aerial mapping, 3D mapping, inspection (railways, buildings, infrastructures), agriculture and security. Each market segment requires very specific technology.’

In terms of the specifications of the cameras used onboard UAVs, Waple said: ‘Weight is a key factor for UAV applications. The FCB EV series fits within the requirements while providing a large feature set, and of course FCB-MA series is even lighter weight (2.2g) which means Sony can offer varied solutions dependent on customer requirements.’ Waple said that, instead of having to be designed for this application, the standard FCB-EV series is used as it is a very flexible product with many different applications, from video security to medical imaging.

The UAV market has remained strong for a while now, but systems are becoming much more widely used as they become more affordable. Waple said: ‘The FCB series of cameras are constantly being developed to incorporate latest Sony imaging technology; FCB modules are used in a wide variety of applications, and the aim is to provide imaging solutions to as many market segments as possible. This of course includes the UAV market, and as the market sees growth we aim to develop more relationships with UAV manufacturers requiring high-quality imaging. FCB-MA13x cameras have been developed to propose a solution when high resolution (13 megapixels), low weight (2.2g) and low power consumption (700mW) is required.’