Biometrics 101: Fingerprint Matching
History of Fingerprinting
Fingerprints are the most common method for criminal identification (and other identification) today. They are useful because fingerprints are unique and don't change as we age. No two people have ever been found to have the same fingerprints. (Even identical twins have different fingerprints.) The ridges on your fingers can be worn down; builders who lay bricks or people who wash dishes will lose some detail. But the ridges grow back. People have tried to remove fingerprints with burns or acid, and that can alter the pattern, but the new pattern is still unique. So that's a pretty handy way to identify people over time.
Before fingerprints were established as a reliable way to identify people, what methods were used? Ancient Romans tattooed people who were conscripted into the army, so that deserters could be identified. In the Middle Ages, branding or maiming was used as a way to mark criminals and warn everyone they encountered to be wary. Hester Prynne was forced to wear a scarlet "A" (for adultery) to humiliate her and identify her to her community in Nathaniel Hawthorne's "The Scarlet Letter". But none of those methods are really great. It's possible to remove the letter and there are other reasons a person might become maimed.
The Bertillionage System
By the 1800s, anthropometry had become the most popular method for criminal identification. Developed by Alphonse Bertillion, the "Bertillionage System" measured 10 specific measurements that were considered to remain constant through adult life. These were height, stretch (left shoulder to middle finger of raised right arm), bust, head length, head width, width of cheeks, length of right ear, left foot, middle finger, and cubit (elbow to middle finger). Photographs of criminals were available, but there was no way to sort through all of the images in a reasonable amount of time. In the Bertillionage System, all of the measurements were put on a card and could be quickly sorted. The system was popular but flawed.
In 1903, the "William West - Will West Case" called the reliability of Bertillon measurements into question. A man named Will West entered a Federal Prison in Leavenworth, Kansas. His face was photographed, and his Bertillion measurements were taken. After this was done, someone noticed that another inmate, known as William West, who was already incarcerated at Leavenworth, had the same name, same Bertillion measurements, and bore a striking resemblance to Will West. Oops! Obviously a better means of identification was needed. (Although it was discovered later that Will & William were identical twins).
In 1892, Sir Francis Galton published a book called "Finger Prints" in which he classified fingerprints based on their main pattern types - arches, loops, or whorls.
The Henry System
Influenced by Sir Galton's book, the Inspector General of the Bengal Police in India, Sir Edward Henry, became interested in using fingerprints to identify criminals. He ordered police to collect fingerprints in addition to Bertillion measurements. He developed a system called The Henry Classification System which allows fingerprints to be categorized by pattern type so that they can be compared. Each finger is assigned a number according to order. The right thumb is number 1 and the left pinky finger is number 10. The pattern type of each finger is determined and then a value for each finger is calculated. Fingers without a whorl pattern are assigned a value of 0. The value of the even numbered fingers are divided by the value of the odd numbered fingers to give a primary grouping ratio. This meant that fingerprints will similar patterns could be stored together and when searching for a matching print, you could eliminate entire categories of prints from consideration, which speeded up the process considerably.
By World War II, the FBI had a collection of over 23 million cards with 400,000 new cards being added each month. By the end of 1943, the FBI employed 13,000 people. All of the cards were stored in filing cabinets and matches were made by humans examining the print and then searching for cards with the matching pattern type. The fingerprint cards used today are largely unchanged from the ones used then.
Development of the AFIS
By the 1960s, the FBI's card collection and the manual classification system was unmanageable. The FBI contracted with the National Institute of Science and Technology (NIST) to study fingerprint automation. By 1969, they were convinced the process could be automated and issued a request for proposals. The first Automated Fingerprint Identification System (AFIS) required an entire room of computers and didn't store fingerprint images. Disk technology was too expensive, so the prints were converted into mathematical templates. In a template, specific features of the fingerprint called minutiae points were mapped. The major minutia features are ridge endings, bifurcations and short ridges. The minutia points are identified and a vector is created from the point to the fingerprint core. A fingerprint template is a mathematical representation of the minutiae points. The main advantage is that templates are much smaller than fingerprint images. A fingerprint template can be as small as 600 bytes and a fingerprint image is around 500,000 bytes. The smaller size means they are faster to match and less expensive to store.
The advantages of the new system were obvious. In 1974, it could take weeks or months to process a fingerprint request - a major problem. The new system could return results the same day. It's important to realize that for law enforcement, an AFIS is a tool to search through millions of fingerprints to eliminate the prints that cannot match. The actual match is still made by a human who examines the entire print image, not just the template. Today the FBI's AFIS contains the prints of over 70 million individuals. The current system is called the Next Generation Identification (NGI) system and in addition to the 70 million fingerprints, it contains palm prints, face and iris images.
Despite the growing use of alternative biometric modalities, such as iris image, face image, gait, keyboard typing style, etc. fingerprints are still the most common, most reliable and easiest biometric in use. For a law enforcement application, where latent fingerprints are often found at crime scenes, they are unbeatable.