-----Original
Message-----
I have been involved with recognising people
by the way they walk – gait biometrics – for many years. Starting in 1994, I
was one of the pioneers of the notion that people can be recognised
automatically by the way they walk and we published many of the earliest
publications in the field together with the first book which describes this new
research area.
It is in this context we studied the videos you
supplied which showed 2 robberies being committed on 2 separate days. For
Camera 1 and Camera 4, given their viewpoint we were unable to use our dynamics
approaches so we manually labelled joint positions and found many instances
where these matched on between different events/days on the same camera. We
have provided you with images showing these matches. In summary of the
analysis,
·
·
Camera
4: we found 35 couples of images (from the two different days) where the
difference in joints' position is lower than 3%
·
·
Camera
1: we found 14 couples of images (from the two different days) where the
difference in joints' position is lower than 5%
Our conclusion is then that
the subject who committed the crimes, featured as running in the data for
Camera 1 and Camera 4, is the same person.
Professor
Mark Nixon BSc PhD FIET CEng FIAPR
From: Mark Nixon [mailto:msn@ecs.soton.ac.uk]
Sent: 30 July 2008 09:41
To: Walker Mick A - QK
Subject: RE: Images
Hi Mick
It’s with my secretary now but
she’s paddled off somewhere. Whilst I wait, how’s this:
Detective Sergeant M. Walker
Priority Crime Unit
Wembley Police Station
Brent
29
July 2008
Dear
Sergeant Walker
I write in connection with the video footage
you supplied recently concerning a theft in London. These videos concerned two
thefts (which we called Camera 1 and Camera 4) and one of a suspect (which we
called Station) waiting in a police station, with Constable Iten.
As you know, we have been involved with
recognising people by the way they walk – gait biometrics – for many years.
Starting in 1994, we were one of the pioneers of the notion that people can be
recognised automatically by the way they walk and we published many of the
earliest publications in the field together with the first book which describes
this new research area. Essentially the automated research is rather more
recent that the notion: Shakespeare mentions several times that people can be
recognised by their gait.
It is in this context we studied the videos
you supplied. For Camera 1 and Camera 4, given their viewpoint we were unable
to use our dynamics approaches so we manually labelled joint positions and
found many instances where these matched on between different events/days on
the same camera. We have provided you with images showing these matches. In
summary of the analysis,
·
Camera
4: we found 35 couples of images (from the two different days) where the
difference in joints' position is lower than 3%
·
Camera
1: we found 14 couples of images (from the two different days) where the
difference in joints' position is lower than 5%
Our conclusion is then that
the subject who committed the crimes, featured as running in the data for
Camera 1 and Camera 4, is the same person.
With the Station data, we are much less sure
of a comparison. The data temporal resolution is too low to allow successive
comparison of images in the sequence with the scene of crime data, in the way
used before. An additional point that might guide us is the height of the
barriers, measured at the centre of the pathway. If you were to have higher
temporal resolution data of the suspect we would be closer, and it would be
better if the subject was running, with a view similar to that of the Camera 1
or Camera 4.
Kind Regards,
Professor
Mark Nixon BSc PhD FIET CEng FIAPR
Professor Mark Nixon
School of Electronics and Computer Science
University of Southampton, SO17 1BJ, UK
Tel: UK (0)23 8059 3542
Fax: UK (0)23 8059 4498
http://www.ecs.soton.ac.uk/~msn
Mark, the fax number is 0208 7333191 but if you could give me a
ring first cos its in another office and could get lost,
-----Original
Message-----
From: Mark Nixon [mailto:msn@ecs.soton.ac.uk]
Sent: 29 July 2008 18:08
To: Walker Mick A - QK
Subject: RE: Images
Hi Mick
Sure, I’ll do it first thing
tomorrow and fax it to you then. What’s the fax number?
Regards, Mark
Professor Mark Nixon
School of Electronics and Computer Science
University of Southampton, SO17 1BJ, UK
Tel: UK (0)23 8059 3542
Fax: UK (0)23 8059 4498
http://www.ecs.soton.ac.uk/~msn
Mark, thanks for this, but could you put it into some sort of
headed report that basically says that you have compared the images from the
the offences on the 2 days and that from your initial investigations there
is a very strong probability that the person committing the offences on the 2
days are indeed the same. Then if you quote the percentages. Also
at the start if you are going to write the report could you put your
qualifications and expertise in the area and how many years experience in gait
analysis etc etc then that would be good. If that could be e mailed or faxed
tomorrow as we are at court on Thursday with it.
Any problems ring me asap
Mick Walker
Detective Sergeant
Priority Crime Unit
Wembley Police Station
Brent
Tel - 43249 - 0208 7333249
Mob - 07737342956
-----Original
Message-----
From: Mark Nixon [mailto:msn@ecs.soton.ac.uk]
Sent: 29 July 2008 16:21
To: Walker Mick A - QK
Cc: mickwalker4655@aol.com;
Iten Carly - QK
Subject: Images
Mike
The comparison files are at http://www.ecs.soton.ac.uk/~msn/comparison.zip
Overall,
- camera 4: has 35 couples
of images (from the two different days) where the difference in joints'
position is lower than 3%
- camera 1: has 14 couples
of images (from the two different days) where the difference in joints'
position is lower than 5%
That looks close indeed. I’m in tomorrow and then out until
Aug 15.
Regards, Mark
Professor
Mark Nixon
School of Electronics and Computer Science
University of Southampton, SO17 1BJ, UK
Tel: UK (0)23 8059 3542
Fax: UK (0)23 8059 4498
http://www.ecs.soton.ac.uk/~msn
**********************************************************************
It is the policy of the MPS that:
MPS personnel (or agents working on behalf of the MPS) must not use
MPS systems to author, transmit or store documents such as electronic mail
(e-mail) messages or attachments:
* containing racist, homophobic,sexist, defamatory, offensive, illegal
or
otherwise inappropriate material;
* containing material requiring a protective marking higher than
RESTRICTED, (and not higher than NOT PROTECTIVELY MARKED across
the internet) without the use of approved encryption;
* containing personal data for use other than in accordance with
the notification(s) under the Data Protection Act, 1998 of the
system(s) from which the data originates.
* This Email message has been scanned for viruses and contents.
**********************************************************************
**********************************************************************
It is the policy of the MPS that:
MPS personnel (or agents working on behalf of the MPS) must not use
MPS systems to author, transmit or store documents such as electronic mail
(e-mail) messages or attachments:
* containing racist, homophobic,sexist, defamatory, offensive, illegal
or
otherwise inappropriate material;
.* containing material requiring a protective marking higher than
RESTRICTED, (and not higher than NOT PROTECTIVELY MARKED across
the internet) without the use of approved encryption;
* containing personal data for use other than in accordance with
the notification(s) under the Data Protection Act, 1998 of the
system(s) from which the data originates.
* This Email message has been scanned for viruses and contents.
**********************************************************************
* This Email message has been scanned for viruses and contents.