[GRASS-dev] Fwd: Proposal for GSoC-2015

[Amitabh]

Hello everyone,

*My proposal:*
I propose to develop a software to mine RSI images' spectral bands
including the visual,ultraviolet and infrared bands.This geospatial mining
will be done at bit level of bands to ensure maximum accuracy.If you look
at the products available in the market they offer predictions
for the entire image instead I would allow users to choose customized areas
to mine upon.

*How this process would work?*

*-->*This software can be used for multiple purposes like in case of Crop
Yield production for farmers.
*-->*The software would need an event(RSI Image of Crop Yield of a
geographical region X) and a contributing factor i.e. a factor that
contributes towards the success of the event like in this case it would be
a RSI Image of rainfall of the same geographical region X.
*-->*After the inputs are fed,then a multimedia data mining algorithm would
find a kind of mapping between rainfall and crop yield.
*-->*Once the mapping is found like Rainfall[ R<180,G>210,B<150 &&
B>125]-->Crop_yield[R>210,G>220,B<40] etc. then the farmer/user can
enter "n" such rainfall RSI images of different geographical images and the
yields over there can be predicted.

*Web application:*
Once I was done with the development of codes.Then I thought of building
the web application for my work and I uploaded the codes on
the server side and then those were executed using shell scripting in PHP.

*Versions of packages:*
1. RGB-->RGB (1 Event and 1 contributing factor)
2.RGB-->GREY (1 Event and 1 contributing factor)
3.RGB-->RGB (One Event and 3 contributing factor)

*Explanation of Codes on GITHUB:*
parm11.java-->Converts RSI images into Band Sequential Format.
parm22.c-->Converts Band Sequential Format into bit Sequential Format.
parm33.c-->Generates Itemsets and rules using Peano count Tree association
rule mining algorithm.
parm44.java-->Using rules and the input generates the predicted image.

*Technical specifications : Java swings,C,PHP,HTML5,CSS3.*

Please provide feedback about the same.I want to add this feature to GRASS
GIS.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.osgeo.org/pipermail/grass-dev/attachments/20150330/e5045070/attachment.html>