Leon Harmon — The Recognition of Faces — block-portraits in Scientific American

1973

Artists: Leon Harmon · Year: 1973 · Medium: Block-portraits — photographs digitally reduced to coarse grayscale arrays · Substrate: Digitized photographs at 16×16, 32×32, and 64×64 resolution; each block averaged to a single grayscale value; printed at large scale · Place: Murray Hill, New Jersey, USA · Institution: Bell Telephone Laboratories Leon Harmon's November 1973 Scientific American article The Recognition of Faces presented a research programme that had been developing at Bell Labs since the Studies in Perception I mosaic seven years earlier. Harmon had been systematically investigating the minimum image information required for face recognition — at what resolution does a face stop being a face? At what spatial-frequency does recognition succeed or fail? His block-portraits — photographs reduced to 16×16 or 32×32 grayscale arrays and printed at large scale — are both research data and striking visual objects in their own right. The most famous of the block-portraits is a 64×64 image of Abraham Lincoln. At the correct viewing distance, the portrait resolves into a recognizable Lincoln; closer up, it disintegrates into grayscale squares; too far and it becomes an abstract field. The image has been reproduced in innumerable psychology textbooks, perceptual-science studies, and pop-cultural contexts since. What Harmon's work demonstrates is that the face is a perceptual object with a characteristic spatial-frequency signature. Recognition is not about resolution; it is about information at specific scales. The argument has shaped face-recognition research for fifty years, including every contemporary machine-learning face-identification system. Harmon is often paired with Knowlton in histories of Bell Labs' computer art, but his individual research contribution on perception was distinct and substantial. Lore The Lincoln block-portrait is a descendant of the Studies in Perception I mosaic in method, but where the 1966 piece combined the perceptual demonstration with Knowlton's artistic collaboration, the block-portraits are Harmon's own research. His Scientific American article was among the most-reproduced pieces the magazine ran in the 1970s, and the Lincoln image in particular has appeared in hundreds of subsequent psychology textbooks. Sources • Harmon, Leon D. — The Recognition of Faces (Scientific American, November 1973) • Bell Labs Technical Memoranda • Computer History Museum — Harmon archive