Take the 2-minute tour ×
Geographic Information Systems Stack Exchange is a question and answer site for cartographers, geographers and GIS professionals. It's 100% free, no registration required.

In order to extract the vegetation mask in ecognition, with object base analysis, with creation of the rule sets, the buildings were also classified! Any idea how to extract building from vegetation? I tried to use shape and also asymmetry, but other ideas are most welcome.

share|improve this question
2  
what software are you using and what is the format of your processed lidar data? –  artwork21 Aug 17 '11 at 13:43
add comment

4 Answers

There are several online videos showing how to use Image Analysis with Lidar data with eCognition.

Image Classification methods are key to extracting the correct data.

http://www.ecognition.com/news/ecognition-tv/

Vegetation (Tree Canopy Extraction) http://www.ecognition.com/news/ecognition-tv/extracting-tree-canopy-lidar

Building Extraction http://www.ecognition.com/news/ecognition-tv/obia-lidar-technical-demo-pt24

share|improve this answer
add comment

There is a solution in LAStools called lasclassify.exe (README) that you could try. The full version (up to a certain point limit) is available on-line for research and evaluation purposes.

share|improve this answer
    
We appreciate replies from vendors and other commercially interested parties, Martin, but in such cases to retain credibility it's crucial that you identify your connection with the solution you are recommending. –  whuber Feb 13 '12 at 15:47
3  
@whuber Martin Isenburg is the developer of LAStools. He is one of (if not the) authority on filtering Lidar data. His algorithms form the basis for many commercial LIDAR software. I wouldn't consider him a vendor, as "his" software is opensource for most uses and is a generous gift to the Lidar community. –  Thad May 7 '12 at 14:00
add comment

If you are using ArcGIS, you could try the LP360 for ArcGIS: Classify tool.

http://www.qcoherent.com/products/classify.html

Free trial

share|improve this answer
add comment

Each Lidar point should have multiple "returns" - the last return is likely to be either building or ground level. Previous ones are likely to be hits on vegetation such as a tree canopy. So my sense is that you could discard all but the last return as being vegetation. That might help.

share|improve this answer
    
your idea is good, but I would take care with high density cover vegetation areas. In this cases some last returns do not reach the ground (besides vegetation, it depends on laser returns density). A solution would be to normalize last returns and filter them above the maximum vegetation height (assuming there are no buildings lower than vegetation). –  Andre Silva Mar 19 '13 at 16:08
    
@Andre Silva, in the case of high density vegetation, one can use "Number of returns" parameter but not last return. The vegetation have more number of returns compared to buildings. –  Sandhya Oct 1 '13 at 12:05
add comment

Your Answer

 
discard

By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.