[OSM-talk] talk Digest, Vol 42, Issue 58

[David Ebling]

>Jo wrote:
>> Robert Vollmert schreef:
>>> I believe gpsbabel supports this kind of
>operation:
>>>
>>> gpsbabel -i gpx -f raw.gpx -x  
>>> radius,exclude,distance=1.5M,lat=30.0,lon=-90.0
->o gpx -F
> filtered.gpx
>>>
>> Of course if you use the actual position of your
>house as the center
> of 
> that radius, it would still be trivial to find
>where you live. That's
> 
> why I tried to exclude points using a polygon
> (saved as a text file).

>> This didn't seem to work though.
>
>There was a presentation by microsoft at google
>concerning managing 
>large groups of anonymous gps traces posted here a
>while back.  Their 
>result for best masking of origin was to choose a
>random point in a 
>circle up to .5km from the origin to use as the
>center of a 1km circle 
>to exclude gave the best results.

Surely with enough tracks you could see the 0.5-1.5km
circles fairly clearly?

How about combining distance from origin with a random
time delay? So take the time of the first point 250m
from the origin, add random period of time between 30s
and 5 minutes before points during which points are
discarded. That way the speed throws in an extra
variable that could only be extrapolated roughly, even
with a large number of tracks. Is this needlessly
complicated though?


      __________________________________________________________
Sent from Yahoo! Mail - a smarter inbox http://uk.mail.yahoo.com