You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
The use case where we saw a problem was in being able to cross streams. An example of the problem scenario is the following. Each cell represents a pixel in a friction surface where "o" symbolizes a good friction value, and "x" symbolizes a nodata value (e.g. impassible).
c1
c2
c3
x
o
o
x
o
The cost distance algorithm prevents travelling between east and west across the nodata pixel. However, it allows travelling diagonally between the "x" pixels. When the o pixels represent a rasterized stream for example, this is not the effect we want. The stream should be untraversable diagonally too.
We should look into modifying the cost distance algorithm to prevent diagonal traversal if the neighboring pixels are nodata values. However, the question is whether we should always do this or control it with a parameter so that caller can tell us what they want. It's possible that there are some friction surfaces where the diagonal traversal should be allowed.
One downside to this approach is that it's possible that a friction surface could have pixels with values like in the above diagram where the diagonal should be passable within the same data set where in other areas they should not be passable - maybe if there are narrow streams in addition to bodies of water that are really close together, or more likely a narrow bridge across water. So this approach could cause problems too...
The text was updated successfully, but these errors were encountered:
The use case where we saw a problem was in being able to cross streams. An example of the problem scenario is the following. Each cell represents a pixel in a friction surface where "o" symbolizes a good friction value, and "x" symbolizes a nodata value (e.g. impassible).
The cost distance algorithm prevents travelling between east and west across the nodata pixel. However, it allows travelling diagonally between the "x" pixels. When the o pixels represent a rasterized stream for example, this is not the effect we want. The stream should be untraversable diagonally too.
We should look into modifying the cost distance algorithm to prevent diagonal traversal if the neighboring pixels are nodata values. However, the question is whether we should always do this or control it with a parameter so that caller can tell us what they want. It's possible that there are some friction surfaces where the diagonal traversal should be allowed.
One downside to this approach is that it's possible that a friction surface could have pixels with values like in the above diagram where the diagonal should be passable within the same data set where in other areas they should not be passable - maybe if there are narrow streams in addition to bodies of water that are really close together, or more likely a narrow bridge across water. So this approach could cause problems too...
The text was updated successfully, but these errors were encountered: