Hi Regan,

If you have your image in ENVI format, then you can load it in MatLab using 'enviread' function:

http://www.mathworks.com/matlabcentral/fileexchange/27172-envi-file-readerwriter/content/enviread.m

In the same way, you can load it using 'multibandread' function.

Check out: http://www.mathworks.es/es/help/matlab/ref/multibandread.html

Hope this helps you!

Hi Regan,

You could do worse than starting with the Landsat data for your area - landsat.usgs.gov . The GLOVIS browser allows you to pan through space and time to see what quality/quantity of imagery is available, plus the possibility to download quicklooks and full data products. The main page has heaps of useful information - e.g. the Landsat Data Users Handbook which guides you through image calibration (which is eminently achievable in Matlab as long as you have sufficient RAM). After that, you'll need to scan the literature for appropriate atmospheric corrections and classification (try an unsupervised cluster analysis for starters - also possible in Matlab - to get a feel for the range of surfaces you're looking at).

Hope that helps,

Jill.

Regan,

Prof. John Jensen of the University of South Carolina is the author of reference courses in remote sensing (http://planet.uwc.ac.za/nisl/Remote_sensing/Jensen/Lectures_RSE/) and digital image processing (http://planet.uwc.ac.za/nisl/Remote_sensing/Jensen/Lectures_DIP/). As a minimum documentation, I suggest you to read these lessons, and afterwards to focus on your subject. In addition, J. Jensen wrote several books you may consult. Also, "Fundamentals of Remote Sensing" tutorial is made available online by the Canada Centre for Remote Sensing.

Concerning the satellite images that may be freely available to you, I agree with Jill's recommendation as regard GLOVIS archive and geoportal (http://glovis.usgs.gov/AboutBrowse.shtml). Look for LANDSAT or ALI images.

Finally, in your place I would choose a dedicated software (IDRISI is very cheap for education; or ENVI, ERDAS, etc.) to concentrate the work to the proper satellite image processing and interpretation. On the other hand, if your purpose for PhD is to develop programs for advanced data processing, matlab is a good choice ... in case you have and know it at a high level.

Thanks to both Jill and Vintila..

Hello Regan,

If you going to use ENVI and Landsat satellite data as mentioned for you then you prefer supervised classification and use maximum likelihood technique which will give you best result in terms of accuracy for classification of image.

Thanks patil. Actually I got image from Bhuvan open data set from NRSC, ISRO. But image is tilted, then I don't know how to read it in matlab.

Regan, I am sorry because I didnt worked in matlab...but you can contact to bhuvan team of NRSC through there contact mail, you will get on NRSC website...you may get some assistance from them.....

Hi Regan,

If you have your image in ENVI format, then you can load it in MatLab using 'enviread' function:

http://www.mathworks.com/matlabcentral/fileexchange/27172-envi-file-readerwriter/content/enviread.m

In the same way, you can load it using 'multibandread' function.

Check out: http://www.mathworks.es/es/help/matlab/ref/multibandread.html

Hope this helps you!

Hi Regan,

Manuel is right - if NRSC have done some ortho/other projection on their imagery, you need to know what that is. I see the website includes handbooks -

http://bhuvan.nrsc.gov.in/bhuvan/IRS_Publications that may include their product specification. I wish the UK had high-res data like this freely available!

Normally, large, global datasets are provided either in netcdf, hdf or grd format, and there are tools (including Matlab) which can read these (netcdf.open, ... hdfread). If you need more help, tell us what type of data you're working with.

Good luck, Jill.

Dear Regan Dhaksnamurthy

Greeting

firstly you must download your intrest area from USGS land sat image from the web sit http://glovis.usgs.gov/ and you must have to register , and the you can download your image and then you must to make atmospheric corrections (convert from DN to Reflectance) and after that you can use several method one of them classification the image and using maximum likelihood or use index of Urban and you can do that in several programe like ERDAS and ENVI and Gis but i prefer ERDAS Imagin.

Thanks. My NRSC data set is geo-referenced and orthorectified by them. I go on it. But let me know about any tutorials/manual. i.e after dataset what about registration or geometric transformation then processing steps.

Jensen's and CCRS tutorials are a very good start, as I mentioned earlier. They are not at an advanced level, but represent an excellent background. You may also look at the David Landgrebe's classical work (https://engineering.purdue.edu/~landgreb/), which takes you from fundamentals to more advanced multi / hyperspectral topics. The latter are valuable for urban applications; you might look in Glovis if there are Hyperion images acquired on your site.

A hybrid method was used to classify buildings in the following publication:

Article Automatic building extraction in dense urban areas through G...