Working with WorldView-2 imagery and decision tree classification method

The Urban vegetation cover is a critical component in urban systems modeling and recent advances in remote sensing technologies can provide detailed estimates of vegetation characteristics.

In this study we will try to classify urban vegetation characteristics, including density condition, using an approach based statistically developed decision trees. This study extracted urban vegetation derived from high spatial resolution WorldView-2 imagery for the City of Depok, West Java, Indonesia.

Pixels with NDVI (Normalized Difference Vegetation Index) values greater than or equal to some values (GE), and less than or equal to (LE) are identified by the following expression:

{ndvi} GE -1.0 LE 0.49 -> No vegetation
{ndvi} GE 0.5 LE 0.69 -> Sparse vegetation
{ndvi} GE 0.7 LE 0.79 -> Moderate vegetation
{ndvi} GE 0.8 -> Dense vegetation

WorldView-2 images of Depok City using band NIR1,Red,Green (RGB-753)
Result of decision tree classification of vegetation density in Depok City.
White: No vegetation (impervious surface, man-made infrastructures, water bodies, cloud cover)
Dark green: Dense vegetation
Light green: Moderate vegetation
Yellow: Sparse vegetation
Advertisements

11 thoughts on “Working with WorldView-2 imagery and decision tree classification method

  1. Salam.
    I have tried the NDVI for WV-2 imagery processing in Erdas but failed.
    However, when image is subset in small image, NDVI can be processed?
    Why?
    Please help me..

  2. The first issue is going to be in your RAM buffer, you need minimum 1 GB RAM for imagery processing, the other issue is likely to be hard disk space, raster operations create temp files in your “workspace” directory that are quite large.

    ERDAS can not handle to big data size, thats why WV-2 images cut into smaller tiles. Please work on this tiles if your study area is to big. using ERDAS-the issue isn’t just the dataset size causing a crash but also the no data areas (null area) not being properly defined

    The solutions:
    1. Rescaling your data from 16bit into 8bit
    2. Compress tiff file to ECW file
    3. Working on tiles to minimize null area (to avoid crash) and mosacking the result then, this the best approach I think

    Good luck! 😉

  3. Salam.
    Dr. If the WV2 imagery was rescaling, can it effect the image?
    thank you very much dr for ur help

  4. Yes, for NDVI technique, transform first and then rescaling
    Rescaling the NDVI (0-200 from -1 to 1) prior to analysis to improve computer visualization

  5. Salam
    Dr, by using WV2 image how can we determine the intertidal region?
    How can we classify with high accuracy eg shallow water, deep water, sandy

  6. I am an Italian student: for my thesis in GIS I need to know the way in which you can transform images WV2 with the method TCT tasselled cap. I ask this because the options that ENVI 4.7 offers are only those for Landsat ETM +. Thank you for your interest ..

  7. Hi,
    For region scale please use band 3 and band 4 of Landsat Images, for my case it produced good result. Good luck!

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s