Fingerprint Analysis: AFIS and IAFIS

Abstract

The Automated Fingerprint Identification System (AFIS) is the primary tool for fingerprint identification used by nearly every law enforcement department in the world. The AFIS can be connected to other databases or they can serve as independent systems which does not communicate with other agencies. The AFIS essentially translates a human image, selects key features, searches the features against the database, and identifies the best potential match from the records. These systems are fast and readily available to law enforcement officers across the United States. The AFIS was developed in response to the United States government quest for quick and accurate biometric identification. This need for prompt identification eventually evolved into the Federal Bureau of Investigation’s (FBI) creation of the nationwide database called the Integrated Automated Fingerprint Identification System (IAFIS) (Komarinski, 2005).

Fingerprint Analysis: AFIS and IAFIS

Introduction

Fingerprint identification is invaluable to criminal investigations and legal processes. Fingerprint evidence can locate, identify, and eliminate suspects, missing persons, as well as human remains (Gaennsslen, 2008). Fingerprints on evidence are fragile and the smallest amount of handling can degrade the quality (“National Institute of Standards,” 2013). Despite this fact, fingerprints remain extremely valuable due to their reliability for identification. Fingerprint identification is often considered the most reliable form of biometrics while the least reliable consist of facial, voice, and hand recognition (Hess and Orthmann, 2010). AFIS technology creates spatial geometry of minutiae of a fingerprint which is then converted into binary code for a searching algorithm (Hess and Orthmann, 2010). A typical fingerprint may contain up to 150 ridges and laboratories will need anywhere from 8 to 15 ridges for matching purposes (Hails, 2014).The AFIS allows examiners to encode minutiae such as ridge endings and bifurcations manually through an auto-extract function. The AFIS will then encode and label the minutiae appropriately (Cole, Welling, Dioso-Villa, and Carpenter, 2008).

History

A German anthropologist, Hermann Welcker, led efforts in the study of friction ridge skin permanence. In 1856, Welcker began printing his right hand and then again in 1897 (Holder, Robinson, Laub, and National Institute of Justice (U.S.), 2011). The use of fingerprinting for purposes of identification was first recorded in the mid-1800s. In 1856, a British officer for the Indian Civil Service, William James Hershel, used fingerprinting to help him identify individuals once he agreed to terms of a contract with them. In 1883, Alphonse Bertillon built a database of criminals utilizing anatomical measurements. In 1905, the United States Army started collecting fingerprints for personal identification wherein other military branches soon followed their example. At the end of World War I, there were almost five million fingerprints collected. 

In 1920, the New York Herald had an article which depicted how criminals could be identified using their fingerprints. In 1922, the Washington Times and the Ogden Standard-Examiner were recommending all babies be fingerprinted and footprinted in order to confirm identities and avoid mix-ups (Bell, 2017). In 1992, the Office of Technology Assessment (OTA) recommended the prioritization of a nationwide AFIS by the FBI and Justice Department agencies. The OTA report discussed the new AFIS technology and described it as a powerful weapon in the war on crime. The OTA called for a new network with the ability to scan, transmit, process, and store fingerprints. The vision of the OTA was a system which provided the foundation of a local, state, and federal partnership for the national exchange of criminal fingerprint data (“FBI Watch,” 1992). In 1999, the FBI initiated the IAFIS which is a database containing nearly 50 million fingerprint records as of 2011 (Saferstein, 2011).

Automated Fingerprint Identification System

The automation process of the AFIS has eliminated the need for a human print examiner to compare two physical fingerprint cards. The searchable database contains fingerprint data or imagery which is collected from a person through hard copy fingerprint cards of electronic fingerprint scanners. The identification process occurs when a fingerprint or prints are searched against a database at a local, state, or federal level. The AFIS utilizes both software and computers to interact with subsystems, including other AFIS’s. Not of AFIS’s are identical and some may or may not link to the state and national database systems (Komarinski, 2005). The AFIS can come in any size as large as the FBI’s IAFIS or as small as local law enforcement agencies. The difficulty exists when these smaller local systems are not linked to the state’s AFIS database due to software configuration issues (Saferstein, 2011).

AFIS technology has the ability to scan and encode fingerprints in order to allow them to be utilized by high speed computer processing. The AFIS has automatic scanning which converts a fingerprint image into digital data showing ridges at their branching points and points of termination. The branching points are known as bifurcations and the termination points are ridge endings. The AFIS also determines the position and orientation of the ridges which allows for storage of each print in the form of a digital geometric pattern (Saferstein, 2011). By 1990, many states in the United States had AFIS’s installed and were using proprietary interfaces. A plan previously approved by the Advisory Policy Board (APB) ID Revitalization Task Force, in 1989, called for the creation of the Integrated Automated Fingerprint Identification System (IAFIS) (Komarinski, 2005).

Integrated Automated Fingerprint Identification System

The largest AFIS system in the US is the IAFIS which is operated by the FBI through their Criminal Justice Information Services (CJIS) division. IAFIS is the conduit for US states to obtain information from other states such as their criminal history and wanted records. IAFIS is central controlled in Clarksburg, West Virginia after previously being located in the J. Edgar Hoover building in Washington, DC (Holder, Robinson, Laub, and National Institute of Justice (U.S.), 2011). The IAFIS has been fully operational since July of 1999. IAFIS is the world’s largest collection of criminal history information and is maintained by the FBI’s Criminal Justice Information Services (CJIS) Division located in Clarksburg, WV (Holder, Robinson, Laub, and National Institute of Justice (U.S.), 2011).

The IAFIS provides automated fingerprint and latent print queries or searches, digital image storage, and digital exchange of fingerprints and responses. These capabilities are available 24 hours a day, 365 days a year. Law enforcement agencies have access to electronic responses from criminal ten print fingerprint submissions within two hours. They have access to responses from civil fingerprint submissions within twenty-four hours (“Federal Bureau of Investigation,” n.d.) Today, the IAFIS has more than 59 million criminal history records and over 210 million criminal arrest cycles. There are 9,000 new records each day, 29,000 file updates each day, and over 104,000 daily fingerprint submissions. The IAFIS provides ten-print services for law enforcement agencies and authorized justice agencies including fingerprint identification and image exchange services.

A ten-print fingerprint submission has ten rolled and flat fingerprint impressions. The ten-print fingerprints are sent to the FBI’s IAFIS after an arrest of employment background check, license, or other purpose. The IAFIS also allows for submission of up to 10 photograph sets for each record. Instructions for submission and retrieval of mugshot photographs are located in the Electronic Biometric Transmission Specification (EBTS) manual. The IAFIS allows for local, state, and federal level searching and comparison of latent fingerprints against the national fingerprint repository. The IAFIS will provide a list of potential candidates with the respective images and other pertinent information which can be used to purposes of comparison (“Federal Bureau of Investigation,” n.d.). The IAFIS often exceeds these standards by completing criminal search requests in under twenty minutes and civil background checks in under three hours. The IAFIS also provides the forensic examiners with an investigative tool which often allows examination of crime scene fingerprints within two hours as opposed to the targeted response time of twenty-four hours (Holder, Robinson, Laub, and National Institute of Justice (U.S.), 2011).

The IAFIS allows fingerprints to be searched even if an individual has a criminal record in another state. For example, the North Carolina State Automated Fingerprint Identification System (SAFIS) communicates and shares information with the IAFIS. The SAFIS computer stores digital data on ten-print fingerprint cards on file at the State Bureau of Investigation (SBI) and the IAFIS computer stores data maintained by the FBI (“North Carolina State Crime,” 2013). It is important to note that the IAFIS does not have an Unsolved Latent File (ULF) of negative results and cannot search palm prints. The IAFIS is also unable to search joints, finger tips, finger sides, or footprints (“North Carolina State Crime,” 2013).

Inherent Risks

Dror and Mnookin (2010) suggest the use of cognitive technologies have increased the sophistication and collaboration between technology and humans and the AFIS is an perfect illustration of this partnership in forensics. Yet, the use and development of cognitive technologies is not as simple as it seems. Dror and Mnookin (2010) claim latent fingerprint identification has been transformed by AFIS technology; however, the human strategies to use the technology have not adjusted adequately enough. Their justification is that the probability of incidental similarity matches or look-alike matches has not been sufficiently explored. They also discuss how the AFIS searches may generate new dangers of bias from forensic examiners.

For example, an examiner may be influenced by where matches appear in the ranking assigned to each of the candidate prints by the AFIS. This may cause humans to examine the potential matches on information other than solely by the objective data available to them (Dror and Mnookin, 2010). According to Alberink, Jongh, and Rodriguez (2014), latent fingerprint examiners make three different types of conclusions during fingerprint comparison including no identification, positive identification, or an exclusion. They argue that positive similarity between fingermarks and fingerprints could be discarded because this process of three category identification is too rigid by only allowing three outcomes (Alberink, Jongh, and Rodriguez, 2014).

Accuracy

Cole, Welling, Dioso-Villa, and Carpenter (2008) address accuracy and validity concerns by showing non-matching images which scored very high on the similarity measure of the AFIS. The study argued that measuring the accuracy and potential for erroneous matches for an AFIS could provide a basis for comparison between automated systems and forensic examiners. Subsequently, this could promote improvements, competition, and innovation in the performance of certain tasks (Cole, Welling, Dioso-Villa, and Carpenter, 2008). Cole, Welling, Dioso-Villa, and Carpenter (2008) found that examiners cannot interpret very high scoring matches in an AFIS as indisputable forensic evidence for a true match.  Although there would be a high number of correct matches, a small number of erroneous attributions would also be made with high confidence.

In the field of fingerprint identification, the AFIS are utilized for automatic searching of candidate sources of fingermarks from crime scenes. Forensic fingermarks are different from fingerprints as they are usually smaller in area size with lower quality and more distortion. This results in a larger variability among fingermarks when compared to fingerprints. The success of finding the source of a fingermark depends on the performance and ability of the technology used for matching. As a result, the system with a high matching performance is often the most desirable (De Jongh and Rodriguez, 2012). Evaluation of fingermark evidence usually includes the comparison of a questioned fingermark to fingerprints retrieved from a database or obtained from a suspect.

Several factors could impact the quality of the fingerprints including smudging, distortion, and the type of background material (Champod, Lennard, Margot, and Stoilovic, 2004). According to De Jongh and Rodriguez (2012), the matching performance of different AFIS technologies can be evaluated with five different performance tests which include testing with different numbers of minutiae, with manual and automatic assignment of minutiae, with originating prints from different regions, with different degrees of distortion, and with different orientations. The study by De Jongh and Rodriguez (2012) found that significant differences in the performances for each test which further emphasized the relevance of developing and applying performance tests on AFIS’s.

Guo, Zhu, and Yin (2018) highlights that unique and stable reference points are important for registration and identification within the AFIS. Most detection methods for reference points scan the fingerprint image pixel by pixel in order to confirm a candidate reference point. This is an inherently complex and time-consuming process. Fingerprint recognition is one of the most reliable biometric identification techniques which is extensively used in security control and personal authentication. There are local and global features of a fingerprint. The local features include the minutia and the global features include the ridge flow pattern. Guo, Zhu, and Yin (2018) examined a strategy to achieve the most efficiency and accuracy of fingerprint detection with reference points. They found that combining the walking algorithm and the EMS-based method, collectively referred to as the WEMS method, allows for both the efficiency of walking algorithm and the accuracy of EMS-based method (Guo, Zhu, and Yin, 2018).

Optimal Database Size

The AFIS databases vary in size and can be as small as a few thousand in local databases. Of course, the IAFIS system is the largest with almost a billion fingerprints (Busey, Silapiruti, and Vanderkolk, 2014). Dror and Mnookin (2010) argue that forensic examiners should pay attention to the size of the fingerprint database due to larger databases having an increased likelihood of producing a similar print which is actually non-matching. They recognize that larger databases increased the likelihood that the questioned print will be in the database; however, larger databases create a greater risk of erroneous identification. The job of the AFIS is to provide similar fingerprints; therefore, by design, it must increase the chances that the forensic examiner will locate similar look-alike prints. This is especially true when compared to fingerprints presented if a subject was identified through traditional techniques other than the AFIS (Dror and Mnookin, 2010).

This produces a potential dilemma for a fingerprint examiner because searching larger AFIS databases may increase their chances of finding someone while also increasing the number of false positives. Dror and Mnookin (2010) suggest chances of finding the suspect tends to asymptote while the chances of a non-matching similar print increases without bound. As a result, in spite of an increase initially, this eventually creates a decrease in the sensitivity of the database as more fingerprints are included. Subsequently, an optimal database size does exist and it should be considered when interpreting matches from a larger database. A database must not become too large to the point where the results of non-matching prints occur too often. One mitigating option for agencies would be to adopt a policy wherein a greater number of minutia should be required before the results from a large database are fully accepted (Busey, Silapiruti, and Vanderkolk, 2014).

The Future

The biometrics industries have progressed since the 9/11 terror attacks and there are signs that the general public is supportive or at least not actively opposed to biometrics in identification (ID) documents. For example, 72 percent of Canadians agree with the concept of introducing a national ID card which require a fingerprint along with a photograph (Lyon, 2008). Biometrics have many applications in today’s technology-based world. For instance, some mobile banking applications utilize biometric modalities to enable a secure payment directly from user smartphones (Blanco-Gonzalo, Lunerti, Sanchez-Reillo, and Guest, 2018). The (AFIS) helps law enforcement officials review fingerprints from individuals they encounter during the conduct of official business. The participation of international law enforcement agencies such as the International Police Organization (INTERPOL) makes it easier to identify an individual through fingerprints (Bell, 2017).

The IAFIS has evolved over time in order to improve the performance and accuracy, adapt to legislative directives, and enhance advanced technological capabilities (Holder, Robinson, Laub, and National Institute of Justice (U.S.), 2011). The popularity and reliance on biometric technologies such as the AFIS will only continue to grow with the advancement and progress of their reliability and validity. Fingerprints may replace standard identification documents such as passports or driver’s licenses in the future. Some airports utilizing fingerprint readers for their security processes in the Transportation Security Administration’s PreCheck program. Major airlines are testing biometric identification systems which may allow passengers to use their fingerprints as in lieu of their standard ticketing or boarding passes. Companies are using fingerprints for employee identification and access to work sites (Bell, 2017). Additionally, the AFIS will continue its migration from forensics o the civil arena (Komarinski, 2005).

Conclusion

The AFIS has changed the identification business model by allowing for identification fingerprint images through ridge characteristics and minutiae. Fingerprint identification is among the most accurate forms of identification available in today’s technology. It isn’t impacted by an individual’s name, gender, or age (Komarinski, 2005).

References

• Alberink, I., Jongh, A., & Rodriguez, C. (2014). Fingermark Evidence Evaluation Based on Automated Fingerprint Identification System Matching Scores: The Effect of Different Types of Conditioning on Likelihood Ratios. Journal of Forensic Sciences, 59(1), 70–81. https://doi-org.ezproxy2.apus.edu/10.1111/1556-4029.12105
• Bell, D. (2017). The History of Fingerprinting. Science Teacher, 84(8), 68. Retrieved from http://search.ebscohost.com.ezproxy2.apus.edu/login.aspx?direct=true&AuthType=ip&db=aph&AN=125936630&site=ehost-live&scope=site
• Blanco-Gonzalo, R., Lunerti, C., Sanchez-Reillo, R., & Guest, R. M. (2018). Biometrics: Accessibility challenge or opportunity? PLoS ONE, 13(3), 1–20. https://doi-org.ezproxy2.apus.edu/10.1371/journal.pone.0194111
• Busey, T., Silapiruti, A., & Vanderkolk, J. (2014). The relation between sensitivity, similar non-matches and database size in fingerprint database searches. Law, Probability & Risk, 13(2), 151–168. https://doi-org.ezproxy2.apus.edu/10.1093/lpr/mgu002
• Champod, C., Lennard, C. J., Margot, P. A., & Stoilovic, M. (2004). Fingerprints and other ridge skin impressions. Boca Raton, FL: CRC Press.
• Cole, S. A., Welling, M., Dioso-Villa, R., & Carpenter, R. (2008). Beyond the individuality of fingerprints: a measure of simulated computer latent print source attribution accuracy. Law, Probability & Risk, 7(3), 165–189. https://doi-org.ezproxy2.apus.edu/10.1093/lpr/mgn004
• De Jongh, A., & Rodriguez, C. M. (2012). Performance Evaluation of Automated Fingerprint Identification Systems for Specific Conditions Observed in Casework Using Simulated Fingermarks. Journal of Forensic Sciences, 57(4), 1075–1081. https://doi-org.ezproxy2.apus.edu/10.1111/j.1556-4029.2012.02114.x
• Dror, I. E., & Mnookin, J. L. (2010). The use of technology in human expert domains: challenges and risks arising from the use of automated fingerprint identification systems in forensic science. Law, Probability & Risk, 9(1), 47–67. https://doi-org.ezproxy2.apus.edu/10.1093/lpr/mgp031
• FBI Watch. (1992). Communications of the ACM, 35(1), 9. Retrieved from http://search.ebscohost.com.ezproxy2.apus.edu/login.aspx?direct=true&AuthType=ip&db=aph&AN=12571675&site=ehost-live&scope=site
• Federal Bureau of Investigation. Integrated Automated Fingerprint Identification System Flyer. Retrieved from https://www.fbi.gov/file-repository/about-us-cjis-fingerprints_biometrics-biometric-center-of-excellences-iafis_0808_one-pager825/view
• Gaensslen, R.E. (2008). Introduction to forensics & criminalistics. McGraw-Hill.
• Guo, X., Zhu, E., & Yin, J. (2018). A fast and accurate method for detecting fingerprint reference point. Neural Computing & Applications, 29(1), 21–31. https://doi-org.ezproxy2.apus.edu/10.1007/s00521-016-2285-9
• Hess, K. M., & Orthmann, C. H. (2010). Criminal investigation (9th ed.). Delmar Cengage Learning.
• Holder, E. H., Robinson, L. O., Laub, J. H., & National Institute of Justice (U.S.). (2011). The fingerprint sourcebook. (pp. 97-264). Washington, DC: U.S. Dept. of Justice, Office of Justice Programs, National Institute of Justice. Retrieved from https://www.ncjrs.gov/pdffiles1/nij/225320.pdf
• Komarinski, P. (2005). Automated fingerprint identification systems (AFIS). 1st ed. Amsterdam: Elsevier Academic.
• Lyon, D. (2008). Biometrics, Identification and Surveillance. Bioethics, 22(9), 499–508. https://doi-org.ezproxy2.apus.edu/10.1111/j.1467-8519.2008.00697.x
• National Institute of Standards and Technology & National Forensic Science Technology Center. (2013). Crime Scene Investigation: A Guide for Law Enforcement. Largo, FL. Retrieved from https://www.nist.gov/sites/default/files/documents/forensics/Crime-Scene-Investigation
• North Carolina State Crime Laboratory (2013). Evidence Guide. Retrieved from https://www.crime-scene-investigator.net/evidence-guide-NC-2013.pdf
• Saferstein, R. (2011). Criminalistics: An introduction to forensic science (10th ed.). Prentice Hall.