postgis/raster/TODO
Bborie Park 9db97233ea Fixed line issues. Added another TODO item.
git-svn-id: http://svn.osgeo.org/postgis/trunk@9997 b70326c6-7e19-0410-871a-916f4a2858ee
2012-06-29 02:06:47 +00:00

68 lines
2.4 KiB
Plaintext

$Id$
This is the TODO list for PostGIS Raster. Items in this TODO file should be
kept brief and to the point. TODO items should only be those items that have
yet to be ticketed in PostGIS trac and lack specifics.
== Simple Projects ==
-- C version of ST_Intersects() in vector-space --
A C version of this function should be faster than the current plpgsql
version.
== Larger Projects ==
-- Spatial Relationships functions --
In addition to the ST_Intersects() function, the following functions should be
implemented.
- ST_Contains
- ST_ContainsProperly
- ST_Covers
- ST_CoveredBy
- ST_Crosses
- ST_Disjoint
- ST_DFullyWithin
- ST_DWithin
- ST_Within
- ST_Overlaps
- ST_Touches
- ST_Within
- ST_Equals (?)
- ST_Relate (?)
- ST_RelateMatch (?)
Like the ST_Intersects() function and its variants, these functions will need
to support both raster/raster and raster/geometry. It may be easier for these
functions to be done in vector-space where the rasters' bands are converted to
pixels and then spatially tested. This should permit the quick delivery of
useful functions though it may not be the fastest solution.
-- n-raster Iterator --
To properly handle aggregate and n-raster mapalgebra operations (mapalgebra,
union, etc), a generic iterator function is needed. This also pushes the
core mapalgebra functionality down to rt_core instead of rt_pg.
-- aggregate Stats functions --
Instead of stats functions using tablename, columnname for a coverage, it
should be using aggregate functions. This change isn't too significant for
most of the stats functions. BUT, the histogram and quantile functions will
need to be rewritten to use 1-pass algorithms that do not require prior
knowledge of the size of the incoming stream. The current coverage quantile
function is a 1-pass algorithm but requires prior knowledge of the size of
the stream.
-- neighborhood tile support --
When processing a tile's pixel in a neighborhood calculation, there needs to
be a way to get the neighboring tiles so as to provide the neighboring pixels
of the pixel being processed. This is a difficult question because rasters
are passed to functions with no information about context. Simply passing
a tablename and raster column is inappropriate as the table and raster column
could be more complex, such as when a table contains daily observations of a
temperature surface.