Rich Helms – CARES

CARES (Computer Assisted Recovery Enhancement System), a computer system for aging missing children announced on July 4, 1986, was a joint project between the Metropolitan Toronto Police and IBM Canada Lab. CARES was the first computer-based child-aging system in the world.

Click to see videos about CARES

The key creators of CARES were Rich Helms and Bill Groves of the IBM Canada Lab and Bette Clarke, Metro Toronto police artist. Bette worked with a Toronto medical research team to learn how people grow and age. Bette then developed growth algorithms that Rich and the IBM team turned into computer image editing software. While the images look primitive by today’s image standards, realize this was 1986, over 25 years ago, and just getting a 640×480 image into a computer was a technical challenge. The PC/AT used was a 8 Mhz 80286 with 2 megs of memory and a pair of 30 meg- hard disks. The image printer in the press conference was a prototype from Mitsubishi that was the state-of-the-art printer at the time.

Rich Helms’ CARES work was recognized with a Metro Toronto Police Appreciation Plaque in December 1992. CARES was featured on Top Cops (US television show) in January 1992.

CARES is still in use at Metro Toronto Police. (Toronto Police Services Forensic Identification Services). The software is now Adobe Photoshop and the hardware is commonly available, but the algorithms and techniques are an evolution of the original work.

There were several interesting findings in the CARES project. First was the use of colour verses grayscale. The computer systems in the 1980s struggled with colour, as did the printers. The IBM colour printer used at the time printed only 8 colours at 72 dpi. The original display at the press conference displayed just 256 colours. The work moved to a Matrox display adapter that showed full 24 bit colour at 640×480, but the team stopped using colour in favor of gray scale. This was not done to avoid technical difficulties, but rather because gray-scale photos were better for recognition. If a colour photo of a person is distributed, people focus on the colour over the features. Use a gray scale photo, and the features become the focus.

How people age was studied in detail at The Hospital for Sick Children and with Dr. Farkas. Thousands of photo collections of people were cataloged and analyzed. This work formed the basis of the CARES algorithms. The reason for the research is interesting. Assume a child is in an automobile accident and goes face first into the windshield, crushing the side of his face. The reconstructed side of his face will not grow, so if you must predict the adult look and reconstruct to that.

There are four characteristics that cannot be predicted though:

  1. Hair colour
  2. Hair style
  3. Weight
  4. Teeth

As all are easily changable (e.g. teeth due to braces), these could only be guesses based on current trends.

“CARES was an amazing part of my life. I worked on some high-profile cases such as enhancing the last photo of Alison Parrot walking past a bank window on the way to being killed. Bette and Bill are now retired while my interest in image and media continues into my current work.”