The NCMEC has a 97% recovery rate of all missing children reported to it, with over 900 safely returned children whose age progressed photos were advertised on TV and on milk cartons. So how do they do it? Turns out there’s a small team of retired forensic detectives using Photoshop and fine art skills to re-imagine what these children might look like as they grow older. 
Joseph Carson was abducted by his non-custodial dad as a toddler and was missing for about five years when a customer at an auto parts store saw that a PSA showing his age progressed image was strikingly similar to a kid who just happened to be in the store at that exact time.
Glen Miller, NCMEC:
I supervise the forensic imaging unit here at the National Center for Missing and Exploited Children. I’m a retired police forensic artist — I’ve been here since 1992. As a detective in the police force, I often created composites from witness memories of bank robbers and rapists. Here at the Center, the emphasis is on aging faces of long-term missing children. It’s different than working from memory, but there isn’t a software available that automatically ages photos.
When a child goes missing, we usually get a photo with the report. As time goes on, though, the photo becomes less and less valuable, especially if the child was very young when he went missing. That’s where age progression comes in. To come up with the best possible progressed image, we begin with a photo of the child and of the biological parent &mdash the father if it’s a little boy, the mother at if it’s a girl — at the age that the child would now be.
My colleague Joe Mullins worked on the most recent image of Jaycee. He had to study 11-year old Jaycee’s face closely, and become familiar with all her unique features — the eyes, the eyelids, the shape of the nose. 80% of likeness is recognizable in the eyes. We’re constantly dealing with the subtleties of aging. What makes someone appear 15 and not 29? He battled that while holding onto the unique facial qualities that set Jaycee apart.
When Jonathan Ortiz was just two years old, his mom ran off with him to Guadalajara Mexico, after attempting to kill his dad by feeding him a milkshake laced with pesticide. She was arrested and extradited back to the US eight years later, but Jonathan wasn’t with her. Here, you can see that the age progressed image of 10-year old Jonathan (middle) created from a photo of him as a one-year old is quite similar to the actual photo taken (far left) when he was found — especially around the mouth.
We use Adobe Photoshop CS4 to manipulate the photos. We stretch the face to approximate growth, blend it with parental photos, and put a hairstyle on each child. The clothes are transformed to be more appropriate for that age. We use powerful Macs with lots of memory and speed, and drawing tablets instead of mouses. With this technology, we can complete one age progression in about three hours.
When we look at the child’s face and family photos, we pretty much know what we’re going to do with it right away. We try to do an age progression every two years until age 18, and then every five years after that. We continue to age progress children unless we’re specifically told not to or until the child is located. Last quarter, we produced 131 age progressions. I enjoy seeing the transformation as I manipulate the photos.
We build faces in virtual environments for people to recognize, but the only way we really know we’re successful is by having results. We can compliment each other on how great an age progressed image is, but the public is the true test of success. To say we love feedback is an understatement. We crave it. It encourages parents of long term missing kids that there’s hope, and that’s one of the most important things about what we do. We’re giving people their identity back.