The Kaplan Daguerreotype dates from the early 1840s and is thought by some to depict the young Abraham Lincoln. Competing authorities, from historians to reconstructive surgeons, have weighed in with their professional opinions as to whether or not the man in the image is the former president. In the early 1970s, a computer program developed at the University of Texas at Austin added its automated determination to the debate. The program did not settle the conversation, however, but instead highlighted just how difficult it was to formalize faces for automated recognition, especially among competing standards of identification. This talk explores that computer program, among the first of its kind, its design, and its place in conversations about automated identification from the daguerreotype debate to the New York State Police Department.
For some forms of pattern recognition like letters and numerals, an automated point-by-point comparison could be relatively successful at the time. This method did not work well for faces, however, because photos of the same person differ a great deal point-by-point depending on factors like head rotation and lighting. The developers at UTA therefore sought to correct photographs for deviations from “face forward” before looking for a match. To do this, they introduced a “Standard Head” whose facial proportions would be assumed of all faces as a starting point from which to measure deviation and correct for rotation and tilt. The program serves as a window into the competing expertise and racial norms that characterized early automated facial recognition.
