matching child fingerprints

[wilbur]

Hello.

I have read that here, in Europe, fingerprints are recorded for all immigrants and even for small children ages 6 and up.
I was wondering how effective is this given that children grow and the electronic fingerprint systems will not be able to match their prints later when they are adults.
In other words, can a match be made mechanically, using a computer, between a template taken at 6 and that person at 20?

thanks

[Ken Chappell]

In theory the size of the fingerprint should not be an issue since the relationship between points would be the same.
Computerized matches have been made on adults with distorted fingerprints from skin flexion and both temporary and permanant scars. As with all fingerprint matches the system works best when good quality know samples are entered into the database and a good quality questioned sample is encoded. The computer tech term for this is G.I.G.O. for Garbage In = Garbage Out.

The computer doesn't "read" a fingerprint the same way that humans do. It reads a mathematical algorithm that looks more like a star constellation than a fingerprint.

[wilbur]

When i saw skin flexion i immediately thought of serious weight change. Is that what you mean? Because probably it would be a similar situation to growing in some sense.
Also, you said a constellation of points is what the computer uses for matching. But if i have a 4x4 meter picture of the night sky and the same picture in my book, how can I match that? I have to see if a star is to the right of the first star, and then if there are 3 stars right above it and so on. I have to keep track of the scale to make sure its the same everywhere... If I don't know the constellation before hand, like i don't know that its the Big Bear constellation, its probably easier to rescale one of the pictures.
Is that (tediously matching each star or rescaling or both) the way computers do it?
sorry for the crude analogy, its the top my understanding

[Graham F]

Wilbur
The relationship between each minutia contained within the finger or palm print does not change. They stay the same, except the fingerprint ridge detail enlarges with the growth of the individual. A child grows in size and so does their hands, but the relationship between the fingerprint minutia, remains unchanged. The detail of a 6 years olds fingerprint will be the same when they are 60. The issue for the mathematical algorithm mentioned by Ken is whether a size relationship between the minutia has been used in the algorithm. This may be a problem in the computerized search method to identify the prints of the 60 year old individual with the prints taken of the same person when they were only 6 years old.

A direct (non computerized AFIS system) manual comparison of the known prints of the same person at 6 & 60 years old should not be an issue and relatively easy, depending on how well the prints were taken. So a match can both be made using a computer, or manually, with a few small caveats.

[wilbur]

thanks for the reply.

I still want to press you a little bit. If not all algorithm compensate for the size relationship why not just store all the images in the same size regardless of their actual size. The image must be 1x1 inch or something. Even if the actual finger is .5 inches.
That's what's giving me a sense of either not understanding or genius.

[kevin]

Wilbur - to add onto what Graham and Ken mentioned I think you are missing that the relationship of minutia to one another is done through intervening/interval ridge counts in addition to (or rather than) spacial differences even in an AFIS search...it really isn't very different from what examiners see every day with distortion. AFIS algorithms can account for spacial 'discrepancy' because they rely on the ridge counts. So weight change, growth, etc will not change the ridge counts of the minutia and their relationship(s) to one another.

[PCC]

My agency entered latents from a 1974 homicide. In 2000 one latent was identified to someone who was 6 years old at the time of the crime. So even though that childs hands and fingers grew the computer was still able to identify the now adult responsible for leaving those latents. And NO that identification did not lead to an arrest.

[josher89]

This is a bit off topic but not completely. The new version of ULW (6.3.6--you can download it from fbibiospecs.org) no longer requires ridge counting as a requirement prior to submitting a latent print. They have also added the ability to search palm prints as well as increasing the penetration level of the search (it used to be 30%; I don't know what it is now but I heard they were shooting for it to be around 70%). So, even though a child grows up, their spatial relationships of minutiae (or ratios of distances between minutiae) don't change.

I printed both of my girls when I could actually begin to see good detail and I will continue to print them until they won't let me anymore. I'm doing it so I can look at the physical distances between minutiae--I know skin's flexible and the ratios won't be exact!--but I'm just curious to see what I'll see. It's another 15 years in the making so stay tuned if you care.

[timbo]

I know our NAFIS in Australia had a 'Juvenile Matcher Algorithm' selection in the search launch screen - I believe designed exactly for this 'spatial' issue to further refine the search. I was not aware of it being used much at all. I had one cold homicide case, about 30 years old, where I identified a print on a plastic lolly/candy counter box through a NAFIS search to a person who had since become a police officer - at the time of the murder, he was 5 or 6 (and of course clearly not involved). I got the hit without using the juvenile matcher algorithm. As a test, I launched the search again with the juvenile matcher algorithm - it very slightly improved the score, but the candidate was still in the same place in the list (3rd or 4th on the list, if memory serves me correctly....)