Hi,
I am in the first experiences with GRASS and I have a problem with the i.maxlik module.
I have to semi-automatically classify two orthophotos already geo-referenced and rectified. As a first operation, I create a group and a subgroup in which I put the two orthophotos. Next, I create a signature with the i.cluster module (initial number of class = 5; iterations = 30; percent convergence = 95%; minimum number of pixels = 5). But in the resulting output there is a problem because the only classified part is the overlap zone between the two orthophotos … why?
I hope I was clear
GC
On Tue, Sep 18, 2018 at 3:38 PM Giuseppe Cillis <giucillis@gmail.com> wrote:
Hi,
I am in the first experiences with GRASS and I have a problem with the i.maxlik module.
I have to semi-automatically classify two orthophotos already geo-referenced and rectified. As a first operation, I create a group and a subgroup in which I put the two orthophotos. Next, I create a signature with the i.cluster module (initial number of class = 5; iterations = 30; percent convergence = 95%; minimum number of pixels = 5). But in the resulting output there is a problem because the only classified part is the overlap zone between the two orthophotos … why?
you need to either classify each orthophoto separately (which is probably not what you want) or first patch (the different bands of) the two orthophotos together with r.patch, then run i.cluster + i.maxlik. When patching the two orthophotos together, make sure the current region covers both orthophotos, e.g.
g.region -p rast=orthophoto1,orthophoto2
Markus M
I hope I was clear
GC
grass-user mailing list
grass-user@lists.osgeo.org
https://lists.osgeo.org/mailman/listinfo/grass-user
Thanks for the message,
but when I try to run i.cluster with just one raster, the module doesn’t work because: " subgroup need two images" .
GC
Il giorno mar 18 set 2018 alle ore 21:34 Markus Metz <markus.metz.giswork@gmail.com> ha scritto:
On Tue, Sep 18, 2018 at 3:38 PM Giuseppe Cillis <giucillis@gmail.com> wrote:
Hi,
I am in the first experiences with GRASS and I have a problem with the i.maxlik module.
I have to semi-automatically classify two orthophotos already geo-referenced and rectified. As a first operation, I create a group and a subgroup in which I put the two orthophotos. Next, I create a signature with the i.cluster module (initial number of class = 5; iterations = 30; percent convergence = 95%; minimum number of pixels = 5). But in the resulting output there is a problem because the only classified part is the overlap zone between the two orthophotos … why?you need to either classify each orthophoto separately (which is probably not what you want) or first patch (the different bands of) the two orthophotos together with r.patch, then run i.cluster + i.maxlik. When patching the two orthophotos together, make sure the current region covers both orthophotos, e.g.
g.region -p rast=orthophoto1,orthophoto2
Markus M
I hope I was clear
GC
grass-user mailing list
grass-user@lists.osgeo.org
https://lists.osgeo.org/mailman/listinfo/grass-user
On Wed, Sep 19, 2018 at 12:43 PM Giuseppe Cillis <giucillis@gmail.com> wrote:
Thanks for the message,
but when I try to run i.cluster with just one raster, the module doesn't work because: " subgroup need two images" .
... this is because it requires two maps or more as input. Do you have
greyscale photos? If yes, i.e. you want to classify a single-band
image, then you can use other options:
- generate synthetic channels (e.g., with r.texture) to be
additionally fed to i.cluster/i.maxlik
- use segmentation, e.g. with i.segment
However, probably you have R, G, B available? Then, as per email by
Markus Metz, you need to first patch channel-wise the two orthophotos
into 3 mosaiks, then feed that to i.cluster/i.maxlik.
HTH,
markusN
I solved with the synthetic channels generation.
Thanks
GC
Il giorno mer 19 set 2018 alle ore 22:39 Markus Neteler <neteler@osgeo.org> ha scritto:
On Wed, Sep 19, 2018 at 12:43 PM Giuseppe Cillis <giucillis@gmail.com> wrote:
Thanks for the message,
but when I try to run i.cluster with just one raster, the module doesn’t work because: " subgroup need two images" .… this is because it requires two maps or more as input. Do you have
greyscale photos? If yes, i.e. you want to classify a single-band
image, then you can use other options:
- generate synthetic channels (e.g., with r.texture) to be
additionally fed to i.cluster/i.maxlik- use segmentation, e.g. with i.segment
However, probably you have R, G, B available? Then, as per email by
Markus Metz, you need to first patch channel-wise the two orthophotos
into 3 mosaiks, then feed that to i.cluster/i.maxlik.HTH,
markusN