[PATCH] Filter and color based on [P|H]DOP

[anonymous]

sometimes we need precise measurements and there should be a way to tell the precision of gps points and maybe filter them.

[h-a-l-9000@…]

I have such a patch, maybe I'll post it some time.

​http://home.arcor.de/h-a-l-9000/img2/dilution.png

[Hasienda]

Please add the patch here. I'd like to use it too, but I'm no Java code writer. With the patch this should be a cheap improvement and make it's way much sooner into JOSM trunk.

HDOP might be enough, but since PDOP is a professional standard, it would be better to have this in place, or even an option to choose a dispersion type, if dataset has multiple of them.

[h-a-l-9000@…]

Isn't PDOP just a combination of HDOP and VDOP? In a 2D-Editor the elevation precision shouldn't be that important...

Adding the patch (diff). The JOSM devs are probably not too happy with it as it is a quick hack I made to be able to map my tracks properly. It only works with GPX files as for NMEA import I use my NMEA improvement patch (adding both patches won't work either as there is a small dependency between the two).

If this does have a future it would be neat if the OSM server provided dilution data for GPS points. Seems not too common in the data of other users though.

[anonymous]

diff patch color-encoded h-dilution

[anonymous]

I don't like the patch as it is.

• Too many changes which are unrelated
• No switch to turn this on/off.

[xeen]

Based on h-a-l-9000's patch

[xeen]

This implements dilution using the same color system that is used by velocity. It also makes coloring work with lines disabled (for all modes) and colors GPS points like it would color the lines (not sure it was intentional to style them differently).

Since I have no GPX file with HDOP information, I couldn't test it properly.

On a side: Is it normal that small GPS points don't render anymore? For some reason drawRect(x, y, 0, 0) seemed to work before but now doesn't. Suggestions welcome.

[stoecker]

A single note after reading patch:

I prefer

..................

.........

compared to

..............

.........

because this make breaking lines more obvious. A normal statement can't start with

or && or +.

Will test that patch later.

[xeen]

Look like trac ate some chars. I believe I know what you meant, here's an updated patch

[h-a-l-9000@…]

Two things:

• The NMEA importer needs to be updated too, I attach the patch.
• My good tracks are purple/blue, the bad ones have red and green in them - doesn't match the description green = good and red = bad.

Otherwise nice work. I'm missing the "legend" though, maybe I'll make that myself. Thinking about an icon in the upper toolbar.

[anonymous]

Replying to h-a-l-9000@arcor.de:

• My good tracks are purple/blue, the bad ones have red and green in them - doesn't match the description green = good and red = bad.

Do you refer to your original patch or my updated patch? I changed the colors to use the red/green scheme in order to avoid having to draw a legend. It's easier to grasp a shade of colors compared to 10 random ones, too. If it draws this "weird" like you explained with my patch, could you please send the file to me so I can test locally? If you have privacy concerns, you can also send it to stefan@… .

[anonymous]

I understand now what is happening - your coloring scheme resembles the Light spectrum with blue and ultraviolet (purple) as the "best" colors.
​http://en.wikipedia.org/wiki/File:Electromagnetic-Spectrum.png

In the display preferences dialog it says red = high, green = low, thus the confusion. Should say blue/purple = low. And no, please don't dumb it down to red-green.

I'm attaching a screenshot with some tracks drawn with the new color scheme.

[h-a-l-9000@…]

Tracks drawn with new color scheme

[anonymous]

In r1425.

[xeen]

We should probably change the wording in the preferences then. Got any good suggestions that make it understandable without having to add a legend? I used red/green because these are pretty much the colors you see everyday when using color by velocity.

If dilution is always pretty low we might want to change it. Currently it is:

Dilution of Position (red = high, green = low, if available)

hmm maybe:

Dilution of Position (red = high, blue = low, if available)

Not sure people understand the light spectrum (red yellow green is pretty obvious, e.g. traffic lights) but blue?