hariasa

Can you be more specific? Obviously, you need to convert these images into something ERDAS can read (this is done in the SNAP toolbox). Do the images open in ArcGIS or QGIS?

The TWOPAC software by DLR can also classify shapes (both pixel classification in raster and shape classification in a shapefile). It is a QGIS extension, though it is a bit hard to find where to download. There are some publications about this software, and maybe you can ask the authors if they have a link to download, since it is open source software. hXXps://www.researchgate.net/publication/225020547_TWOPAC_-_A_new_approach_for_automated_classification_of_satellite_imagery

I know that with eCognition you can do such a thing from a raster (raster shapes, make polygons, classify polygons based on many different attributes), you probably could do it with a shapefile too, not sure though. A free alternative of eCognition is Orfeo, but I do not know if you can use that to do what you want.

ESRI is just a greedy corporation that wants to monopolize the sector with stupid proprietary formats, incompatibility with open standards and crappy software. They buy companies, and if that isn't possible, make their own mediocre product which everyone will use because it is integrated and because of name building. They don't care about the consumer. They don't care about GIS. All they care is cold hard cash. They are just a bunch of Ferengi. But not only ESRI is like this, also ERDAS, Microsoft, Apple, Samsung.... It's just how the world is.

I have tried this with GM 15, and it downloads the data, but it only creates blank rasters

Hi, I want to convert data from a WMS (all aerial photos) to offline rasters. Is this possible? Sure you can load WMS directly in your GIS application, but this is not suitable for an internet connection as slow as mine. Programming examples would be very helpful. Thanks

This solution does not work. I have added the registry key with the Chinese Simplified codepage (936), and when saving the shapefile after editing all chinese characters revert to gobbledygook. Step 1: Did not work for me, same issues as above, all characters after editing revert to nonsense. Step 2: Sounds interesting! Hopefully it will get back to you because this issue is really annoying. Why doesn't ArcGIS simply accept CPG files??? Thanks for helping though, if you or anyone else has more ideas, let me know P.S.: by the way, to translate a DBF from a shapefile more efficiently, open in OpenOffice, select appropriate Codepage, Save as XLS format, upload to Google Documents, make a new column in Google spreadsheets, apply a formula such as =GoogleTranslate(<cell-to-be-translated>,"de","en") and drag down, download as XLS format, save as DBF format replacing your existing DBF file. Note that the column name should be of the same format as the other column names or it won't work! (Do this with open office because Excel does not allow codepages to be set). Long procedure, but if you have like 50.000 entries it's much faster than translating every single one of them manually...

I have a shapefile with Chinese characters in it. When I edit the shapefile and put an english translation, the chinese characters dissappear because of the encoding my system uses. I could change the system locale, but this is to me a last resort. Is there a way to force ArcGIS to read things in a certain encoding? (for example messing with the DBF header?) QGIS demonstrates how it is done; it just accepts CPG files. Thanks!

Really sad. Landsat 5 was a great satellite. Hopefully Landsat 8 will have a long and productive life

Copy cloud services now has a special free give away for online storage. You get 15 GB for free, + 5GB for each referral. What makes Copy different from most other cloud services (such as Skydrive and Dropbox) is that it has no limitation on file size. So perhaps it is interesting for remote sensing purposes/syncing of smaller projects, since those images can sometimes be too large to be accepted by other services. Also the pricing is attractive, if you are into that (group users etc.) See the link below: https://copy.com?r=AMB6mp

Hi all, I was just wondering if someone has used WEKA (or RapidMiner) to classify RS data. Searching on the internet revealed some papers about people using it, but with only for very small areas. Does anyone have experience with large amounts of data to classify with either of these programs? Since it seems kind of impossible to convert a large multi-band raster to the silly CSV/ARFF format WEKA uses. Thanks

Since I wasn't that familiar with ENVI (actually studying it for my project) I found that FME was the easiest to configure.

Hi all, I have a lot of rasters of the same size, location, projection, file type etc. that I need to stack. Ordinarily I would do this with the Stack Layer option in ERDAS, but now I have to do it a whopping 60 times * 9 bands. There must be a better way to do this, so I thought of FME. But FME with the rasterbandcombiner baffles me. I insert 9 rasters in there and I get 9 rasters back, although I wanted 1 raster with 9 bands back instead. I know you can do it in ArcGIS too, but my classifier (Randomforest in R) does not take kindly to rasters that are created with ArcGIS (no matter what format). Does anyone know how to stack these rasters together in a more automated fashion (in another program or using the rasterbandcombiner in FME)? Also, the rasters are not necessarily all the same bit range (8-bit etc.) and not necessarily of the same spatial resolution. Much thanks! I already found the solution, you are supposed to use the sort. FME can only stack similar bands, not like ERDAS stack anything you like. If someone has an automated solution for that last bit, that would be great. Thanks all.

I really don't like the ERDAS vegetation index calculator thingy. So what I did instead is program a python script to calculate all the indices i need automatically from multiband optical imagery.

Eventually, I performed the RandomForest method with R. It was quite easy to program, with the help of a few tutorials on the R help. I have not tried EnMAP yet, I will do so now and post here how it went. Having a lot of variables in R with very big imagery does not always go well... sometimes randomforest in R 'forgets' to classify certain things the first time. It's also quite RAM intensive, so if anyone else wants to do it, you need at least 4 GB of RAM to do a decent random forest classification with R for a decent image. Thanks nomlas for your help! +1