Back to Blog & Articles Page >>


This article was written for and first appeared on July 14, 2016.  To read the original CLICK HERE

July 14, 2016

Image Analysis:  The Reel and the Real

It was waiting on my desk when I got to the office, a shiny CD containing three images from a string of unsolved mail thefts. It’s just one of hundreds of cases involving surveillance images that I’ve worked over the years, and when I opened the files it was pretty much what I expected: grainy images of a vehicle too far away and displaying too few details.

If my life were a primetime crime drama, I could simply push a button and each image would slowly come into focus on my monitor, magically finding nonexistent pixels that form into a suddenly legible license plate number. Better yet, I could snag an image of the suspect as he hands a stolen credit card to the clerk in exchange for five packs of cigarettes and 10 lotto tickets, and by pushing the same button I could zoom in on his open wallet and read the name off his exposed driver’s license.

That would be the reel world: the world of film and TV and Hollywood’s limitless imagination.

In the real world of crime analysis, there are no magic buttons, no shortcuts. It’s rare when a surveillance camera is close enough, positioned correctly and of high enough quality to capture a usable license plate number. In most cases you get a blob of pixels, and when you zoom in you get a bigger blob of pixels. Sure, you can enhance the image to some degree, but nothing like you see in the movies.


About eight years ago, after an unsolved armed robbery, my frustration with poor-quality surveillance images led me to do what crime analysts do best: improvise. I developed a database-driven analytical system called Forensic Vehicle Analysis, which allows me to take dark or fuzzy images and generate a list of vehicle models and years that might be a possible match based on over 50 factors, such as taillight orientation or license plate position. Since that time, I’ve used Forensic Vehicle Analysis on cases that range from petty theft all the way up to kidnapping, rape and homicide. 

As for the CD on my desk and the unsolved string of mail thefts, I was able to analyze the images and identify the vehicle as a black 2000-2002 Saturn L-series sedan. After checking the Department of Licensing database for similar vehicles registered in the county, I found a couple dozen that matched the color. I immediately recognized that one of the vehicles was registered to a known mail thief we had sent to prison a few years earlier. Turns out the suspect had been released from prison right around the time the latest string of mail thefts began.

As the case wrapped up, mail from over a hundred victims was recovered, including driver’s licenses, vehicle titles, credit cards, debit cards, bank records and legal paperwork. 

Maybe someday the real world will catch up to the reel world. Until then, we do the best with what we have.




Back to Blog & Articles Page >>