One of the unfortunate effects of my visual impairment is an inability to perceive people’s physical details in a timely manner. I don’t know how anyone else matches individuals within their visual or auditory fields to memorized patterns, or how quickly the process takes place, so I can only explain how I think I do it and solicit responses from others who’ve considered the matter, or even conducted research.
Since it is nearly impossible for me to pick out details such as eye colour, small formations such as wrinkles, or small pigmentations, beyond a few centimetres’ distance, my initial pattern match tends to rely upon overall outline; in other words, I first recognize people by silhouette, followed by easily-resolved features such as hairstyle, facial hair, or large tattoos.
Because I rely so heavily upon a person’s shape for initial recognition, and I share the common tendency to look at someone’s head for this purpose, haircuts and hats can cause my recognition process to fail on the first pass. More than once, people I would otherwise immediately recognize and greet have spoken to me for up to thirty seconds before I realize who they are, due to changed hair lengths, styles, or headwear. Voice and (relatively) fine features failed to trigger a match until the subject said something that made me realize I should know that individual.
Strangely, for someone who often has to rely upon auditory sensory capability a bit more than usual, particularly for work, matches by voice take a moment longer than I would like, and I’ve found myself occasionally unable to correctly match voices to memories. I’m also mostly incapable of reproducing sounds with my voice. I don’t have a natural talent for performing music, nor did I put enough effort into acquiring that capability at an early age to overcome that lack of innate ability. I suspect these quirks are related.
tl;dr: neurobiological pattern matching is funky shit.