[GRASS-user] Output of i.segment

[Jaya Krishnan]

I think I found a solution though. Used i.segment and then used the
segments to make training sites and signatures for performing supervised
classification.

Jaya
<https://ca.linkedin.com/in/jayaakrish>

On 3 April 2016 at 10:07, Jaya Krishnan <jayaakrish at gmail.com> wrote:

> Thanks! However I don't completely understand. Are you saying that there
> is no currently way to classify directly from raster result of i.segment
> based on some raster statistics, and possibly choosing vector training
> areas from this?
>
> Jaya
>
>
>
> On 3 April 2016 at 08:31, Moritz Lennert <mlennert at club.worldonline.be>
> wrote:
>
>> On 31/03/16 01:17, Jaya Krishnan wrote:
>>
>>> Hi fellow GRASS users,
>>>
>>> I have outputs of i.segment (object segmentation) for 30m Landsat
>>> imagery from 4 dates. The segments are around 10000 each for each
>>> imagery. How can I use these outputs for land cover change analysis? Do
>>> I have to vectorize these and recode the classes, or is there a more
>>> efficient way?
>>>
>> >
>>
>> The classical approach would be:
>>
>> 1) segmentation
>> 2) caracterization of the segments
>> 3) selection of training segments or training points
>> 4) classification
>>
>> Currently, in GRASS you have the following addons for the respective
>> steps:
>>
>> 2) v.stats / i.segment.stats
>> 4) v.class.ml / v.class.mlpy / v.class.mlR
>>
>> You can check the section on Object-based classification in [1] for more
>> info (although partially outdated).
>>
>> Moritz
>>
>>
>> [1] https://grasswiki.osgeo.org/wiki/Image_classification
>>
>>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.osgeo.org/pipermail/grass-user/attachments/20160403/0bc03acf/attachment.html>