<div dir="ltr"><div><div><div><div>Hey,<br><br></div><div>For instance in every Lidar point cloud, the order is paramount, because the sensing is structured (lines usually).<br></div><div>So ordering can give you neighbourhood relation for free.<br></div><div>Also, ordering may be important when you want to improve your point cloud registration .<br></div><div>I don't mean point cloud global geolocalisation, but each point localisation in the sensor reference system.<br></div><div>Typically
points are sensed in the sensor reference system, then knowing the
sensor trajectory in WGS, you put the point in WGS (for instance).<br></div><div>Sadly sensor trakectory is always false to a point. Correcting it allow to re-register the points and improve points precision.<br></div><div><br></div>Sorry my answer was a little cryptic. <br><br></div><div>(I talk about LOD in [this presentation](<a href="https://github.com/Remi-C/Postgres_Day_2014_10_RemiC/raw/master/presentation/A%20PostgreSQL%20Server%20for%20Point%20Cloud%20Storage%20and%20Processing.pdf" target="_blank">https://github.com/Remi-C/Postgres_Day_2014_10_RemiC/raw/master/presentation/A%20PostgreSQL%20Server%20for%20Point%20Cloud%20Storage%20and%20Processing.pdf</a>) , p49 and 50)<br><br></div><div>For
ordering points I build an octree, then for each level of the octree (0
to N), I take the point closest to the center of the cell.<br></div>This gives you at most 8^K points at the level K. <br></div>(here illustration in 2D on real data. Biggest point = most important point = lower Level point. )<br></div>I
don't build an octree in fact, because you can get what point goes in
which cell simply by looking at the binary coordinates of the points
(after offset and scale).<br><div><br>This is cool if the user read the whole level. <br>But you can do something better by allowing continuous-LOD.<br>In this scenario the user can read only a part of a level, and we still want that he gets well spaced points.</div><div><br>For instance the level 2 would have 64 points, maybe the user read points 1 to 30.<br></div><div><br></div><div>If you order points within a level by morton curve, the user will only get points in the upper part of the patch (for instance).<br></div><div>So the user would have no points in the lower part of the patch !<br><br></div><div>We want to do precisely the opposite, and that the user gets points well spaced in the patch. <br>I used random because it was easy, but it is not deterministic.<br>The
inverse morton code provides a cool order to dispatch points and kinda
solve this problem (as well as being easy & fast to write in low
level langages).<br></div><div>We could do more subtle I'm sure.<br></div><div><br></div><div>Cheers</div></div><div class="gmail_extra"><br><div class="gmail_quote">2015-04-17 13:07 GMT+02:00 Peter van Oosterom <span dir="ltr"><<a href="mailto:P.J.M.vanOosterom@tudelft.nl" target="_blank">P.J.M.vanOosterom@tudelft.nl</a>></span>:<br><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">
<div bgcolor="#FFFFFF" text="#000000">
<div>Hi Rémi,<br>
<br>
Good to hear that other persons find our work of interest. Thanks!<br>
<br>
Fair remark about the fact that original order of points can be
significant.<br>
(could you give one example where this is really clear?)<br>
<br>
I do not understand what you mean with 'for LOD I considered using
an inverse morton curve'.<br>
Options for selecting points for higher LODs could be random (not
so smart, but easy) or some spatial analysis (figure our relative
relevance/ importance of points, smarter, but non-trivial). So, in
general two nearly equal (very close) points should not both go to
higher LOD, so better look for the different point to push to
higher level. You could do this with real distance in 3D space,
but it may be easier with distance of 1D morton code. So, when
points are processed (in original sequence), you compute their
morton code, and when next point is significantly different wrt
morton, it goes up (and otherwise stay at lowest level). Is this
what you mean with 'inverse morton for LOD'?<br>
<br>
We really need LOD and smart data organization for efficient use,
current test of 640 billion points we need it (and there is a 30
trillion data set announced for Netherlands, which we also would
like to use efficiently in interactive manner).<br>
<br>
Kind regards, Peter.<div><div class="h5"><br>
<br>
On 17-4-2015 12:35, Rémi Cura wrote:<br>
</div></div></div><div><div class="h5">
<blockquote type="cite">
<div dir="ltr">
<div>
<div>
<div>
<div>
<div>Hey Oscar (and Peter), <br>
</div>
great series of articles, the benchmark was really
interesting !<br>
<br>
</div>
I personally am adamantly against reordering the points
for storage purpose, because the order can have a strong
meaning (physical or theoretical).<br>
<br>
</div>
For instance I re-ordered the points of my patch so that the
points follows a LOD schema.<br>
</div>
So reading more and more points gives a better and better
approximation of the patch.<br>
</div>
<div>This gives free LOD facilities when all the patch have been
re-ordered (free CPU wise, and no data duplication).<br>
</div>
<div><br>
</div>
<div>Now I have to admit that efficient storage and LOD are in
conflict, because storing is about putting the similar points
together somehow, and LOD is about the opposite (kind of
sequential vs random access).<br>
</div>
<div>This has a strong impact, for instance reordering patches
randomly will greatly increase patch storage size in
pgpointcloud.<br>
</div>
<div>It is funny because for LOD I considered using an inverse
morton curve (interleave X Y Z , then read is backward ! ).<br>
</div>
<div><br>
</div>
<div>I'm well aware that we could simply store the original
order in an attribute, and so compression based on morton on
XYZ might be transparent for the user.<br>
</div>
<div><br>
</div>
<div>Anyway, just to say that pgpointcloud is ultimately not
about storing points, but easing point cloud usage in my
opinion.<br>
</div>
<div><br>
</div>
<div>Cheers,<br>
</div>
<div>Rémi-C<br>
</div>
<div><br>
</div>
<br>
<div>
<div><br>
</div>
</div>
</div>
<div class="gmail_extra"><br>
<div class="gmail_quote">2015-04-17 11:02 GMT+02:00 Oscar
Martinez Rubi <span dir="ltr"><<a href="mailto:o.martinezrubi@tudelft.nl" target="_blank">o.martinezrubi@tudelft.nl</a>></span>:<br>
<blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">
<div bgcolor="#FFFFFF" text="#000000"> Hi,<br>
<br>
About the XYZ binding for better compression. In our
research in the NL escience center and TU Delft we have
been thinking (not testing yet though) about one possible
approach for this.<br>
<br>
It is based on using space filling curves. So, once you
have the points that go in a block you could compute the
morton/hilbert code of the XYZ. Since all the points are
close together such codes will be extremely similar, so
one could store only the increments which could fit in
many few bits. We have not tested or compared this with
any of the other compressions but we just wanted to share
it with you just in case you find it useful!<br>
<br>
An additional improvement would be to sort the points
within the blocks according to the morton code. Then, when
doing crop/filter operations in the blocks one can use the
morton codes for the queries similarly to what we
presented in our papers with the flat table (without
blocks), I attach one of them (see section 5.2). In a
nutshell: You convert the query region into a set of
quadtree/octree nodes which can be also converted to
morton code ranges (thanks to relation between
morton/hilbert curve and a quadtree/octree). You scale
down the ranges to increments (like you did when storing
the point of the block) and then you simply do range
queries in sorted data with a binary algorithm. In this
way you avoid the decompression of the morton code for
most of the block. This filtering is equivalent to a bbox
filter so it still requires a point in polygon check for
some of the points.<br>
<br>
Kind Regards,<br>
<br>
Oscar.
<div>
<div><br>
<br>
<br>
<div>On 16-04-15 18:15, Rémi Cura wrote:<br>
</div>
</div>
</div>
<blockquote type="cite">
<div>
<div>
<div dir="ltr">
<div><span style="font-family:monospace,monospace">epic
fail ! I had avoided html just for you<br>
</span></div>
<div><span style="font-family:monospace,monospace"><br>
Dataset |subset size | compressing |
decompressing |<br>
|(Million pts)|(Million
pts/s)|(Million pts/s)|<br>
Lidar | 473.3 | 4,49 |
4,67 |<br>
</span></div>
<span style="font-family:monospace,monospace">21-atributes
| 105.7 | 1,11 | 2,62 |<br>
</span>
<div>
<div><span style="font-family:monospace,monospace">Stereo
| 70 | 2,44 |
7,38 |<br>
<br>
</span></div>
<div><span style="font-family:monospace,monospace">Cheers<br>
</span></div>
</div>
</div>
<div class="gmail_extra"><br>
<div class="gmail_quote">2015-04-16 17:42
GMT+02:00 Sandro Santilli <span dir="ltr"><<a href="mailto:strk@keybit.net" target="_blank">strk@keybit.net</a>></span>:<br>
<blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex"><span>On Thu, Apr 16,
2015 at 05:30:12PM +0200, Rémi Cura wrote:<br>
> OUps<br>
><br>
> Dataset | subset size(Million
pts) | compressing (Million pts/s) |<br>
> decompressing (Million pts/s)<br>
> Lidar | 473.3
| 4,49<br>
> | __4,67__<br>
> 21 attributes | 105.7
|<br>
> 1,11 |
2,62<br>
> Stereo | 70
| 2,44<br>
> | 7,38<br>
<br>
</span>These tables aren't really readable
here.<br>
Could you make sure to use a fixed-width font
to write those tables<br>
and to keep lines within 70 columns at most ?<br>
<br>
--strk;<br>
</blockquote>
</div>
<br>
</div>
<br>
<fieldset></fieldset>
<br>
</div>
</div>
<span>
<pre>_______________________________________________
pgpointcloud mailing list
<a href="mailto:pgpointcloud@lists.osgeo.org" target="_blank">pgpointcloud@lists.osgeo.org</a>
<a href="http://lists.osgeo.org/cgi-bin/mailman/listinfo/pgpointcloud" target="_blank">http://lists.osgeo.org/cgi-bin/mailman/listinfo/pgpointcloud</a></pre>
</span></blockquote>
<br>
</div>
</blockquote>
</div>
<br>
</div>
</blockquote>
<br>
<br>
</div></div><span class="HOEnZb"><font color="#888888"><pre cols="72">--
Peter van Oosterom <a href="mailto:P.J.M.vanOosterom@tudelft.nl" target="_blank">P.J.M.vanOosterom@tudelft.nl</a>
Section GIS technology (room 00-west-520) Department OTB
Faculty of Architecture and the Built Environment, TU Delft
tel (+31) 15 2786950 Julianalaan 134, 2628 BL Delft, NL
fax (+31) 15 2784422 P.O. Box 5043, 2600 GA Delft, NL
<a href="http://geomatics.tudelft.nl" target="_blank">http://geomatics.tudelft.nl</a> MSc Geomatics
<a href="http://www.msc-gima.nl" target="_blank">http://www.msc-gima.nl</a> MSc GIMA (Geo-Info Management&Appl)
<a href="http://www.gdmc.nl" target="_blank">http://www.gdmc.nl</a>
</pre>
</font></span></div>
</blockquote></div><br></div>