Is RTree updated automatically in Boost geometry?
Dear all,
I am now using Rtree in boost geometry.
I contained pointers of spatial geometry data into RTree, whose coordinates are changed during computations.
Here, I have a question. Is rtree updated automatically when the coordinates are changed? Or, we have to reconstruct rtree each time?
I think constructions of the Rtree take much time so that I would like to reduce the lengthy time. If you know some know-how, I would appreciate it if you could tell me.
Kind regards,
Mitsuru
I am now using Rtree in boost geometry.
I contained pointers of spatial geometry data into RTree, whose coordinates are changed during computations.
Here, I have a question. Is rtree updated automatically when the coordinates are changed? Or, we have to reconstruct rtree each time?
I think constructions of the Rtree take much time so that I would like to reduce the lengthy time. If you know some know-how, I would appreciate it if you could tell me.
Kind regards,
Mitsuru
I did not use rtree but I don't think that there is any kind of automatic update. It wouldn't even help since it would make things slower. I.e. how would they check for a change?
Hi,
Instead of changing coordinates, I guess you could remove then insert again with the new values, that way the load balancing can fix up the node.
Instead of changing coordinates, I guess you could remove then insert again with the new values, that way the load balancing can fix up the node.
Dear coder777 and TheldeasMan, and all
Thank you for your reply.
Yes, probably it makes things slower, and I checked the computing time by some timer library.
I will try remove an element and insert one of changed coordinates.
By the way, some spatial indexing techniques (c.f. flann, BVH, Rtree) seem time-consuming for construction of indexing trees.
Nowadays, there appear many applications using real-time spatial indexing (c.f. 3D sensors, AR, computer visions, etc), which requires reconstructions of the trees since coordinates of point clouds etc are updated each time.
For such a real-time spatial indexing, are there suitable algorithms for spatial indexing?
(Perhaps, full construction each time is not done, but partial or high-level-nodes updates are adopted)
If you know, I would appreciate it if you could help me.
Kind regards
Thank you for your reply.
Yes, probably it makes things slower, and I checked the computing time by some timer library.
I will try remove an element and insert one of changed coordinates.
By the way, some spatial indexing techniques (c.f. flann, BVH, Rtree) seem time-consuming for construction of indexing trees.
Nowadays, there appear many applications using real-time spatial indexing (c.f. 3D sensors, AR, computer visions, etc), which requires reconstructions of the trees since coordinates of point clouds etc are updated each time.
For such a real-time spatial indexing, are there suitable algorithms for spatial indexing?
(Perhaps, full construction each time is not done, but partial or high-level-nodes updates are adopted)
If you know, I would appreciate it if you could help me.
Kind regards
Topic archived. No new replies allowed.