
Log Polar Sampling
Hi all,
I'm trying to understand how logpolar sampling works.
By "logpolar sampling" I mean that I have a set of 2D points (x,y),
for each point I need to build a signature which depends on
the "logpolar" distribution of the other points relatively to this point.
The signature is built by taking a pixel (x,y), normalize all the other points
relatively to it ( making (x,y) the origin of the axis ), find the polar
representation of the other points, and sort them in bins.
The value of bins for the point is the signature.
This image may explains it better
http://www.eecs.berkeley.edu/Researc...in_diagram.jpg
Now, my problem is that I don't understand how to decide the maximium
radius. This image represent the polar tranformation, and the number of
bins (R and W)
http://www.cogs.susx.ac.uk/users/dav...lines/fig4.gif
It was taken from here.
Does anyone familiar with this field?
Thanks & Good day.