Bug 231 - [SECTStudio-Classifier] Feature Type Raster Classification not valid
: [SECTStudio-Classifier] Feature Type Raster Classification not valid
Status: NEW
: MapLab
SECT (Studio)
: 3.0
: PC Windows 2000
: P2 normal
: ---
Assigned To:
:
:
:
:
:
  Show dependency treegraph
 
Reported: 2004-02-19 15:51 by
Modified: 2004-02-23 20:18 (History)


Attachments


Note

You need to log in before you can comment on or make changes to this bug.


Description From 2004-02-19 15:51:22
I am unable to classify a raster. Is this even implemented? If not will it be in
the near future? If not than this should be removed form the Feature type list.
Or you could have it in the list but set to inactive, greyed out.
------- Comment #1 From 2004-02-23 12:26:56 -------
it is implemented in Studio but support for generating SLDs with raster
classifications is not done in MapServer so it currently does nothing.

Reassigning to Assefa for comment on when generateSLD will support rasters.
------- Comment #2 From 2004-02-23 13:08:31 -------
There is still something unclear for me in the specs. Here is how the SLD looks
like 

<ColorMapEntry color="#00ff00" quantity="-500"/>
<ColorMapEntry color="#00fa00" quantity="-417"/>
<ColorMapEntry color="#14f500" quantity="-333"/>
<ColorMapEntry color="#28f502" quantity="-250"/>
<ColorMapEntry color="#3cf505" quantity="-167"/>
<ColorMapEntry color="#50f50a" quantity="-83"/>
<ColorMapEntry color="#64f014" quantity="-1"/>

Here is what is specified in the specs :

For example, a DEM raster giving elevations in meters above sea level can be
translated to a colored image with a ColorMap. The quantity attributes of a
color-map are used for translating between numeric matrixes and color rasters
and the ColorMap entries should be in order of increasing numeric quantity so
that intermediate numeric values can be matched to a color (or be interpolated
between two colors).

 It seems that the specs expect to work on range of values for classification
and not on descrete values. (In the example above we would have pixel >= 500 and
pixel < 417 as a first class) There were a couple of e-mails sent to the ogc
mailing lists but no answer to clarify this.

 It  is not a big deal to generate the sld but we have to make a descision on
this issue.
------- Comment #3 From 2004-02-23 13:26:36 -------
This decision will only influence e.g. the look of a legend, not the way the 
colors will be assigned.

The user should be able to influence which label he wants for the first class:
1) "-500 to -417"
2) "-500" 
The second option applies if his data are discrete values.

The spec has the "label" attribute for this which lets the user influence the 
label for the class.

<xs:attributename="label"type="xs:string"/>
------- Comment #4 From 2004-02-23 20:18:49 -------
um ... I think the issue is what to do with a given mapserver class when
generating the SLD document so that the local and remote treatment of the data
is consistent with what the user would expect.  It also has to reflect what we
are capable of representing in our end-user tool, Studio.

My interpretation is that for raster classification, every class should be
treated as a [field] >= value in Studio and then when generating ColorMap
entries for SLD, they can be created with a quantity that reflects the value in
the expression.  I think complex expressions and expressions that do not operate
on the classitem should be ignored or skipped as invalid (for SLD generation).

Studio will need slight modifications to support this, but it will be pretty minor.