Jump to content

Land use classification


ransara

Recommended Posts

yes, you can, if you have the eCognition software (you can find it here on this forum).

eCognition creates images segments (trees, buildings, roads or water bodies) but even so, it will be hard to make a very good clasification map, because your quickbird image has only 3 layers (green, blue, red).

For example, if you have an aditional layer, let's say infrared (NIR) you can use this to better classify vegetation vs buildings or water ! ;)

So, good luck my friend !

  • Like 2
Link to comment
Share on other sites

VietCong, my friend !

you can download eCognition software from this forum. Just give a search on the forum !

About classification techniques using eCognition, there are a lot of information on eCognition comunity or on the eCognition web-page.

Also you can find some interesting articles about land use classification with eCognition on internet (Google).

There are some tips and tricks using this software, but you can find documentetion on most of the links that I have mentioned above ^_^

So, good luck and good work ! :D

Vietnam = nice and interesting country,

VietCong = nice computer game ! :lol:

pac-pac !!

  • Like 1
Link to comment
Share on other sites

ransara you can learn more about basic Remote Sensing just by a simple "search" in this forum, and you will find here a lot of books and information. Also, you can search on Google.

You have an aerial foto (orthophoto) of a city and you whant to make a land use classification ? How big is the image ?

Even with the most powerfull image classification software that you have ( eCognition ! :D ) you will still have some problems making an automated clasification of the aerial photo. But you can use eCognition segmentation tool to create a layer of shapefile polyons of the contours of the buildings, roads, water bodies, etc on your aerial image and then do a manual calssification (in ArcGIS or in eCognition), if your image is not too big and you have a lot of time to work on it.

Good luck, my friend ! ;)

  • Like 1
Link to comment
Share on other sites

  • 2 months later...

Hello Buddy,

Image classification is furely based on objects that you need to classify, henceforth the requirement is is that object sensitive to the band that you have considered.

For example vegeation has higher reflectance in NIR band and water absorbs, so to classify vegetation and water separately you can use and NIR band, similarly for other land uses

Link to comment
Share on other sites

bharathaithal

when you have just a simple aerial image (ortophoto) with only 3 bands in the visible spectrum (red, green, blue) you have limited possibilities to obtain a detailed and accurete classification. For example, if you don't have at least 1 NIR bad, you can't make a good distinction beetwen a river and a forest, using the traditional pixel-based image classification. It will be a major problem with the spectral mixture, most of the classes will have very similar specral responses, so the best way is to use an Object-based image classification method > eCognition software. Then, using eCognition, you can classify the river using not only it's spectral caracteristics, but also other caracteristics (lenght, widith, area, etc) versus the vegetation, at different levels of image segmentation.

So, in conclusion, when you have only a simple aerial photo (3 visible bands) and you want to obtain a detailed classification, you ca't do it without auxiliar information (cadastral maps, LIDAR data, aditional infrared bands), even when you use an Object-based image classification. The traditional pixel-based image classification is excluded from the start for this job !

Good luck!

Edited by Arhanghelul
  • Like 2
Link to comment
Share on other sites

I totally agree with Arhanghelul. I couldn't have said it better :)

eCognition is from what I hear indeed the best or most advanced software to do a object-based classification.

OB classification makes use of contextual information extracted from the objects determined on spectral similarities.

Some years ago I performed such a classification with Spring, which is free Brazilian software.

Lurker did point us to the software already 3 years ago (time flies):

http://www.gisarea.com/index.php?/topic/119-spring-gis-513/page__hl__spring

Current version is 5.2.

Good luck!

  • Like 1
Link to comment
Share on other sites

Me too. I like Ecognition software because it classification by object oriented. It provide many threshold and algorithm to classification. But in Ecognition that make process tree is difficult. it request experience and creativeness.

Edited by HTC
  • Like 1
Link to comment
Share on other sites

  • 2 weeks later...

Dear my friends!

i ask you about: how to smoth polygon when to classify. When i classify object in Ecognition. The object have boudary isnt smooth. The shape is jagged. How to do smooth objects? Thank you very much.

I guess you mean the shape of the objects. It depends mostly on the raster and rules you use. Try high res image.

OT- check out the advancements in imagine objective. i see 2013 brings few new rules in the party. :)

  • Like 1
Link to comment
Share on other sites

What else @rahmansunbeam? The object is jagged when i classify image in Ecognition or Envi. I want to do the object is smooth. If i used toolbox smoth polygon in Arcgis. The object is not exactly. if someone know how to smooth object when to classify, please share these experiences.

Regard!

Link to comment
Share on other sites

What else @rahmansunbeam? The object is jagged when i classify image in Ecognition or Envi. I want to do the object is smooth. If i used toolbox smoth polygon in Arcgis. The object is not exactly. if someone know how to smooth object when to classify, please share these experiences.

Regard!

What segmentation rules you used? the denser the segmentation, the higher the block number, and the cpu usage is higher. On other hand, the higher the resolution, the curves are smoother. My suggestions will be,

-use multiple segmentation techniques (checker board+quad tree+multires...) for better object size with optimum cpu. For really big image i'll take hundreds of dices.

-edit objects if necessary

-output vector then edit

I guess smooth edge like vectors cannot be achieved in raster. If you somehow get to do that, you'll loose lots of pixel from the edge. Using subpixel operation in this case may be clever, but will still be a challenge for the cpu.

  • Like 1
Link to comment
Share on other sites

Yes. I used Multiresolution segmentation. I want to smooth objects because when i print map in scale max (1/1000, 1/500... or more than). The boudary is jagged, it is not powerful. I have mean to use simply polygon or smooth polygon vector in Arcgis to smooth object. Thank rahmansunbeam!

Link to comment
Share on other sites

Yes. I used Multiresolution segmentation. I want to smooth objects because when i print map in scale max (1/1000, 1/500... or more than). The boudary is jagged, it is not powerful. I have mean to use simply polygon or smooth polygon vector in Arcgis to smooth object. Thank rahmansunbeam!

Now i understand. The eCognition outputs are only for analysis, not presentation. To create a map you'll surely need to use vector layers. In this case use AG. For best result use Illustrator.

Link to comment
Share on other sites

Yes, exactly. But i used Toolbox: Polygon smooth or simply polygon to smooth it. The result not perfect. If someone know to make perfect map from the classification result, please share these expenrience!

@rahmansubeam. I dont use Illustrator software. :). I use Arcgis, Mapinfo to make Map.

Edited by HTC
Link to comment
Share on other sites

Yes, exactly. But i used Toolbox: Polygon smooth or simply polygon to smooth it. The result not perfect. If someone know to make perfect map from the classification result, please share these expenrience!

@rahmansubeam. I dont use Illustrator software. :). I use Arcgis, Mapinfo to make Map.

why the polygon smooth is not satisfactory? play a little with the environment settings, threshold and xy tolerance to fine tune the output in certain scale.

(the finest maps we know are tuned within Illustrator before the final printout. AG has some technical limitation to output the finest details in an image, but illustrator is always the best!)

  • Like 1
Link to comment
Share on other sites

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.

Guest
Reply to this topic...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.

×
×
  • Create New...

Important Information

By using this site, you agree to our Terms of Use.

Disable-Adblock.png

 

If you enjoy our contents, support us by Disable ads Blocker or add GIS-area to your ads blocker whitelist