Easy to Implement Improvement to Track Simplification

[Rod Moyer]

TopoFusion's website describes this software's process for "Simplifying" a GPS track as follows:  "The point which when deleted will yield the smallest error from the original track is selected and deleted. This process continues until the desired number of points is reached."

As an engineer, a pretty fair mathemetician, and a computer programmer, I say this is a far more useful track simplification process than offered by some other software, where a track is simplified by merely eliminating points according to their time or distance proximities.

One very easy to implement addition would make this process significantly even more useful: a user option to select the largest acceptable error from the original track rather than the desired number of points after simplification.  This would be easy to implement because the algorithm would not change, only the program's test for process completion would change.  In other words, instead of stopping when the desired number of points has been achieved, the process would stop when the next logical point removal would cause an error, from the original track, greater than the acceptable error.

[ScottMorris]

This is a good idea... I like it.

What sort of uses for this type of simplification do you envision?

[Rod Moyer]

The use for this feature improvement would be no different than for the current Simplify Track feature, to reduce memory usage in a limited GPS unit, and improve performance even where momory may not be an issue.

The benefit of this feature improvement would be that the user would not have to guess how many points a track can be reduced to without seriously impacting its shape & accuracy.  For instance, if we have track of 1000 points, with a great many very tight curves, such as mountain 'hairpin' curves, a simplification to 200 points might make some of the tight curves 'short circuit' and disappear.  A relatively very straight track of 1000 points might be simplified to 200 points without such damage to its shape.

Assume that a user simplified each of the two example tracks using my method, and specifying a maximum acceptable error of 4 meters from the original track.  The very curvy track might be reduced to 500 points and the relatively straight track to 100 points, while both of them maintained their shapes & accuracies to the same acceptable level.

[wayne]

I'm not certain how you would determine what the acceptable error on a track reduction would be.   I ran some experiments a while ago, and put the results here.

http://www.topofusion.com/forum/index.php?topic=3934

[Rod Moyer]

The acceptable error would be up to user judgement based on:

1.  the faintness of the trail (smaller error means less chance of losing it), and

2.  the tightness of curves, especially switchbacks (if adjacent trail segments in a switchback are closer together than the allowable error, the switchback will entirely disappear when the track is reduced - not good).

For example, with the new option I'm advocating, if I can see that adjacent trail segments in a switchback are as close as 5 meters apart, I could set the acceptable error to 3 meters and know that the switchback would not disappear.  Using only the current option of how many points to reduce a track to, it is much harder to guess what that number should be.