Kristin Cowalcijk created SEDONA-361:
----------------------------------------
Summary: Add RS_MapAlgebra for performing map algebra operations
using simple expressions
Key: SEDONA-361
URL: https://issues.apache.org/jira/browse/SEDONA-361
Project: Apache Sedona
Issue Type: New Feature
Reporter: Kristin Cowalcijk
Fix For: 1.5.0
We'll add a new function {{RS_MapAlgebra}} for performing map algebra much
easier. The {{RS_MapAlgebra(rast, pixelType, script, [noDataValue])}} function
takes three to four arguments:
* {{{}rast{}}}: The raster to apply the map algebra expression to.
* {{{}pixelType{}}}: The data type of the output raster. This can be one of
`D` (double), `F` (float), `I` (integer), `S` (short), `US` (unsigned short) or
`B` (byte). If specified `NULL`, the output raster will have the same data type
as the input raster.
* {{{}script{}}}: The map algebra script.
* {{{}noDataValue{}}}: (Optional) The nodata value of the output raster.
The map algebra script will be powered by
[Jiffle|https://github.com/geosolutions-it/jai-ext/wiki/Jiffle]. It is feature
rich and very efficient, since it compiles Jiffle script to java bytecode.
For example, NDVI calculation can be implemented using {{RS_MapAlgebra}} in
this way:
{code:sql}
SELECT RS_MapAlgebra(rast, 'D', 'out = (rast[3] - rast[0]) / (rast[3] +
rast[0]);') as ndvi FROM raster_table
{code}
Users can use any bands in the input raster, this is quite different from the
map algebra function provided by PostGIS. The same NDVI calculation can be
implemented using array-based map algebra functions already existing in Sedona,
but much more verbose and error-prone.
{code:sql}
SELECT RS_Divide(
RS_Subtract(RS_BandAsArray(rast, 1), RS_BandAsArray(rast, 4)),
RS_Add(RS_BandAsArray(rast, 1), RS_BandAsArray(rast, 4))) as ndvi FROM
raster_table
{code}
Currently, we'll only accept one raster as input. Since we can refer to
multiple bands of the raster in the Jiffle script so it won't be a big blocker
for implementing most of the real-world map algebra operations.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
