It's probably also worth mentioning Jon Kerridge has done work in this
area. I believe he's mainly using IR these days, but has also done some
semi-automated pedestrian path logging from cameras. His publications
list is at:
http://www.soc.napier.ac.uk/publication/op/peoplepublications/peopleid/51
In particular, see the recent Willis et al paper in Environment and
Planning B.
Alasdair
Andrew Smith wrote:
> Rachna Gupta asked:
>
>
>>Could anyone please direct me towards work that has been using cameras
>>or other video data to capture movement patterns/ numbers in crowd
>>areas?
>
>
> Transport for London has paid for two studies into this, both managed by
> the Central London Partnership and with my involvement each time.
>
> One finished last year, and tested cctv and infra-red from Springboard
> and Footfall. You can download the report from
> http://www.c-london.co.uk/files/pdf/Report%20-%20Automatic%20Pedestrian%20Counting%20Trial.pdf
>
> At the time of that study, the counting worked ok, but there was no
> useful capture of movement. However, the study prompted several
> manufacturers to put more resources into research and development: hence
> the second study.
>
> That one is currently in the field, and is testing cctv with more recent
> software (to count, to map pedestrian trajectories), and
> vertically-scanning laser radar [aka lidar] (to count). This should
> report later this year. Hitachi and Shibasaki labs in Japan have used
> the laser radar with horizontal scanning, to produce a remarkably
> effective pedestrian-trajectory mapper. However, we are still awaiting
> a commercial implementation of this.
>
> Regards,
> Andrew Smith
--
Alasdair Turner
Course Director
MSc Adaptive Architecture and Computation
Bartlett School of Graduate Studies
UCL Gower Street LONDON WC1E 6BT
http://www.aac.bartlett.ucl.ac.uk/
|