On 2020-12-01 11:14, Warin wrote:
> The differences are less than 10m. (The points of the green track are where
> data exists, the straight lines between those points simply connect the
> measured points. )
>
> The 'simplify way' in JOSM is normal set for a maximum difference of 3m as a
> way of reducing data bloat while sacrificing some accuracy.
What accuracy is optimal for OSM? Why should we sacrifice any accuracy
at all? Who chose 3m as a tolerance figure? That sounds rather high to
me - it's the width of a small road, or half a house. If we are going to
draw a line at all, I would go for something <= 1m.
Truly geometrically redundant points such as intermediate points on a
straight line are fair game, but de-wiggling by applying DP-type
algorithms should be done very sparingly IMHO. Smoothing and curve
fitting may also be fine on GPS traces, as you could consider them as
actually improving the data, but the argument of "data bloat" should not
be invoked as an excuse to arbitrarily sacrifice accuracy.
_______________________________________________
Tagging mailing list
Tagging@openstreetmap.org
https://lists.openstreetmap.org/listinfo/tagging
