Facial Recognition: Technology Rooted in History

December 6, 2017

By: Fardin Barekat, BASc

The process of seeing a person and knowing who they are, commonly referred to as identification, is a non-trivial process which is taken for granted amongst humans, having adopted this skill naturally along their evolutionary process. Throughout human history, this ability to identify other people has been employed indirectly, for example through portraits, and has been done so for activities ranging from the arts, to entertainment all the way through to law enforcement. The twenty first century stands at the precipice, in danger of becoming a cold world dominated by the spectre of biometric invasion and not only lacking personal privacy, but even denying private identity. Among other things, it is the growing trend of the proliferation of social media sources and Internet of Things devices that is blamed for this steady erosion of personal privacy by such practices as biometric identification. It will, therefore, come as a surprise to most people that the procedural use of facial recognition precedes not only social media, but further the internet, both world wars, and even the invention of the programmable computer.

One crucial step towards facial recognition was the invention of the photograph, in the early 1800’s, and then the subsequent photographing of criminals beginning in the 1840s. By 1888 with the standardization of this latter photographing, the now famous Mug Shot was created, enabling law enforcement personnel to turn their efforts into interpretation and identification of mug shots.

An early form of procedure based facial recognition was the Tableau synoptique des traits physionomiques system invented by Alphonse Bertillon in XXXX. This system, most often used by local law enforcement officers, employed a cheat sheet with a variety of uniquely indexed facial features.

Although the Tableau System aided law enforcement activities by creating a common facial code language, by virtue of relying solely on the efforts of examiners, it was wholly non-automated. In 1966 Woody Bledsoe, Helen Chan Wolf, and Charles Bisson created a program which calculated the distances between facial features (ie. Eyes, ears, nose, etc.) which was then compared to reference data. The three pioneers had created the world’s first facial recognition program. Their program, however, still relied on an examiner for manual input of the facial feature locations. This requirement, thereby, limited their program to only being a semi-automated facial recognition program.

Although improvements were made to this program in the 70’s – Goldstein, Harmon, and Lesk employed 21 specific facial features, Kirby and Sirovich employed linear component analysis – these programs still required manual input and, thus, required human interaction. It was not until 1991 that the processing of detecting faces, face detection, was developed. With face detection, the examiner was no longer required for detection of faces, and thus both feature extraction and facial recognition were both fully automated. Therefore, in 1991 the first fully automated facial recognition program was born. And the rest, as they say, is history.

Prilyx Research and Development is at the forefront of biometrics research and integration by developing new techniques and algorithms and integrating with other security related systems.

© 2018 Prilyx Research & Development Ltd. | 2220 – 1050 W Pender St Vancouver, BC. Canada, V6E 3S7
+1 604. 428. 5030 | info@prilyx.com