Monday, June 25, 2012

Mobile Device Remote Identity Proofing Part 3 - Apples to Oranges

Download PDF of complete paper

IV. Apples to Oranges:   

Can a camera in a smart phone be used to capture the necessary images, to include those used for biometric identification, required for the enrollment and subsequent vetting of an individual in an Identity Management System (IDMS)?  Smart phone manufacturers are equipping their newest products with cameras capable of ten or more megapixels with Nokia’s latest offering claiming a forty plus megapixel camera!  This paper proposes using the camera to capture all of the required components to establish and vet an identity so it is important to understand some of the terminology involved.  

Contrary to popular belief more megapixels do not make for a better image.  It is important to understand what makes up a good image and how it is defined within the multiple industries involved. Most people base image quality on the output / final product, the best example being print media.  So this is where we are going to start.

Pictures are printed in DPI or Dots per Inch. For example a newspaper image is printed at 200 to 250 DPI, A magazine image is 400-600 DPI, yet a billboard is typically 30 dpi.  When you print a photo on your desktop printer the optimal setting is for 250 DPI.  Don’t be fooled by the fact your typical desktop printer is capable of far greater resolution, typically from 720 to 1440 dpi. The printer may be able to print very small dots but it can only accurately reproduce colors by combining a large number of dots to emulate various tints. That is why a 250 dpi image offers perfect output quality on a 1000+ dpi printer.  

PPI is Pixels per Inch.  PPI is the resolution terminology used in the Standards promulgated by the American National Standards Institute (ANSI) and the National Institute for Standards and Technology (NIST).  Within the context of this paper PPI is used to define the resolution of the scanning mechanism used to capture a fingerprint.  PPI is an appropriate term to describe scanner input and it is the term used by the applicable Federal standards, but technically, samples per inch (SPI) is more accurate. “For example, if you scan at 200% at 300 PPI or if you scan at 100% at 600 PPI, the scanner [sees] the same data.  The PPI is different for each file, but the sampling of the original by the scanner is the same.  Maximum SPI of a given device is the optical resolution at 100% “(Creamer, 2006)  

How do dots per inch equate to pixels?  The term pixel is predominantly used to describe the digital resolution on monitors, televisions, and smart phones.  A pixel is one dot of information in a digital photograph. Digital photos today are made up of millions of tiny pixel/dots (Mega = Million).  A digital photo that is made up of 15 megapixels is physically larger than a digital photo made up of 1.5 megapixels, not clearer or sharper.  The notable difference is in file size, not picture quality.   If you print a 250 DPI picture on an 8.5 by 11 piece of paper you will be printing a maximum of 2125 by 2750 pixels. Most computer screens display at 100 DPI.  A 1280 by 1024 resolution on your monitor equates to 1310720 pixels or 1.3 megapixels.  This begs the question, why do you need a ten plus megapixel camera to capture a very high quality image?  The answer is you do not.  

V.  Camera Technology

With an explanation of some of the terminology behind us we can explore the use of a digital camera or variant, for the capture of the necessary data for enrollment in an identity management system.  When the FIPS 201 standard was first published capturing a facial image of an individual required, by standard, the use of a three point five megapixel camera.  This level of resolution was at the top end of the capabilities of digital cameras readily available to the public at the time.  Costs in excess of a thousand or more dollars a for a camera meeting FIPS requirements were not uncommon.  That same Camera was also unable to do anything more than capture an individual’s picture.  Today native resolutions on smart phone integrated cameras are commonly five times the historical benchmark.  Exponential improvements in the image capture hardware, firmware and supporting software should also enable these same devices to not only capture a photo but be multi purposed for barcode reading, OCR enabled document capture, Fingerprint image capture, and even iris image capture.  4G and LTE networks now make it possible for high speed efficient exchange of data with next generation networks coming on line reinforcing and bolstering the capability.  Consistent with Moore’s Law the capability of cell phones is on the steep end of the climb with exponential growth and improvements in power, processors, and memory.  

“A digital camera can capture data based on the mega-pixel ability of its CCD.  For example, a 2 megapixel digital camera shoots at approximately 1600x1200. 1600 pixels times 1200 pixels = 1,920,000 total pixels (rounded up)  Usually the camera images have no resolution assigned to them (although some cameras can do this)  When you open a file into an image editing program such as Photoshop, a resolution HAS  to be assigned to the file.  Most programs, including Photoshop, use 72 PPI as a default resolution. (Creamer, 2006)

VI. Establishing ownership

Biometrics is the science and technology of measuring and analyzing biological data.  Biometric identifiers are the distinctive, measurable characteristics used to identify individuals. (Jain, Hong, & Pankanti, 2000) The two categories of biometric identifiers include physiological and behavioral characteristics. (Jain, Flynn, & Ross, 2008)  Physiological characteristics are related to the shape of the body, and include but are not limited to: fingerprint, face recognition, DNA, palm print, hand geometry, iris recognition (which has largely replaced retina), and odor/scent.  Behavioral characteristics are related to the behavior of a person, including but not limited to: typing rhythm, gait, and voice.  

The most common biometric identifiers currently used in IdM systems are fingerprint and facial recognition.  With the current PIV and PIV-I programs a dual approach in accordance with NIST recommendations (NIST, 2003)is used.  The capture of these biometric identifiers is easily within the scope of commonly available commercial technologies incorporated into today’s smart devices.  It is the analogous algorithms required for image analysis and development of minutia for analytical and comparison purposes that pose the challenge.  Current facial recognition software is more than capable of effectively using images captured within the common 8-14 megapixel range of the average smart phone.  The technology is rapidly outpacing the market’s ability to sustain new releases and/or uses as evidenced by Nokia’s release of a smart phone with a 41 megapixel camera sensor dubbed the 808 PureView (Foresman, 2012)  So the specific challenge relates to the fingerprint.


Works Cited

Creamer, D. (2006). Understanding Resolution and the meaning of DPI, PPI, SPI, & LPI. Retrieved May 30, 2012, from

Foresman, C. (2012, March 2). Innovation or hype? Ars examines Nokia's 41 megapixel smartphone camera. Retrieved March 5, 2012, from arc technica:

Jain, A. K., Flynn, P., & Ross, A. A. (2008). Handbook of Biometrics. New York, NY, USA: Springer Publishing Company.

Jain, A., Hong, L., & Pankanti, S. (2000, February). BIOMETRIC IDENTIFICATION. (W. Sipser, Ed.) COMMUNICATIONS OF THE ACM , 43, pp. p. 91-98.

NIST. (2003, February 11). Both Fingerprints, Facial Recognition Needed to Protect U.S. Borders. Retrieved March 5, 2012, from NIST; Public and Business Affairs:

No comments:

Post a Comment

Please keep your comments specific to the topic. Requests for a specific topic can be emailed to