Geoscience Reference
In-Depth Information
we will use the command line tool, otbcli_Segmentation to segment part of a
LISS-III scene and demonstrate the impact of the different segmentation algorithms
(meanshift, edison); additional segmentation algorithms are included in the module
that include: cc, watershed and mprofiles. The segmentation results can either be
output as a raster or vector dataset. First, we present the module documentation that
includes the different flags and options.
-progress
<
>
boolean
Report progress
-in
<
string
>
Input Image (mandatory)
-filter
<
Segmentation algorithm [meanshift/edison/cc/watershed/mprofiles]
(mandatory, default value is meanshift)
-filter.meanshift.spatialr
<
string
>
Spatial radius (mandatory, default value is 5)
-filter.meanshift.ranger
<
int 32
>
Range radius (mandatory, default value is 15)
-filter.meanshift.thres
<
float
>
Mode convergence threshold (mandatory, default value is 0.1)
-filter.meanshift.maxiter
<
float
>
Maximum number of iterations (mandatory, default value is 100)
-filter.meanshift.minsize
<
int 32
>
Minimum region size (mandatory, default value is 100)
-filter.edison.spatialr
<
int 32
>
Spatial radius (mandatory, default value is 5)
-filter.edison.ranger
<
int 32
>
Range radius (mandatory, default value is 15)
-filter.edison.minsize
<
float
>
Minimum region size (mandatory, default value is 100)
-filter.edison.scale
<
int 32
>
>
Scale factor (mandatory, default value is 1)
-filter.cc.expr
<
float
Condition (mandatory)
-filter.watershed.threshold
<
string
>
Depth Threshold (mandatory, default value is 0.01)
-filter.watershed.level
<
float
>
Flood Level (mandatory, default value is 0.1)
-filter.mprofiles.size
<
float
>
Profile Size (mandatory, default value is 5)
-filter.mprofiles.start
<
int 32
>
Initial radius (mandatory, default value is 1)
-filter.mprofiles.step
<
int 32
>
int 32
>
Radius step. (mandatory, default value is 1)
 
Search WWH ::




Custom Search