Jan 032016
 

It can be a murky game comparing photos visually to see if there are resemblances. 

The results are open to individual interpretation and this can be coloured by the wishes and hopes of the person doing the reviewing. Add in to the mix images taken over different time frames such as fifty years apart  for an individual (where  the bone structure should be the the same) and comparing photos of people from different generations in the hope of seeing a resemblance.

In some families there can be a lot of resemblance but there are also many families where if you were not aware of the family structure and just looked for resemblances you would swear they could not be siblings at all. Also be careful of the “Humphrey nose’ “Smith ears” as they could also be seen in other people.

To do a facial comparison you need to look at a minimum of the:

Shape of the face: round, square, oval etc
Eyes: their shape, colour (often we don’t have that information) size, distance between them, position
Nose: shape , size position
Nostrils: shape, size position 
Moles: placement (in modern times often removed)
Teeth: only seen if smiling. In times of bad dental care teeth may also be false or decayed

Remember you also need to being aware of changes due to weight gain/loss and age complications such as losing teeth.

If you can get the images the same size and orientation (and unfortunately we can’t go back in history and ask the person to stand facing the camera please!) using Photoshop or other imaging software you can make one image transparent then overlay it on the other which can be useful for an individual at different ages.

Other software such as Picasa claim to do photo recognition and yes with modern photos I have seen a number of photos identified correctly BUT I have also seen many mis-identifications including a man identified as a woman (an unrelated one at that), different individuals being called the same person etc. The more known photos of that person at different ages helps increase the success rate of identifying other photos but it is not 100% or even 80%.
Even the software used by governments using complex mathematical algorithms on a minimum of eighty points is not 100% perfect. There is an interesting document on the US Federal Bureau of Investigation site which goes into the history of facial recognition and the development of standards and ongoing work to improve facial recognition.



There is more photo matching software becoming available. Lisa Louise-Cooke recently mentioned this one which a Microsoft one listed as  more of a game: Twins or Not so I thought I’d give it a try. You upload two photos to compare.

So I picked one of my mother and myself at around a similar age.















Then I found another  of Mum where she was smiling and made her head a similar size to mine to see if it would be called differently.


Now I admit Mum and I have many similarities but 100%?

So what about me aged differently?
 



Apparently I didn’t age that well. I suspect the glasses have influenced the results. (Changing the order of the upload didn’t change the result)

Then how about Mum and older me?


Interesting result.

I am not saying you shouldn’t look at photographs but I am suggesting strongly that you should interpret with caution especially solely on visual comparisons or even computer comparisons (unless you have access to the FBI, Interpol or other law enforcement software) and even then there are caveats on the results.