Meeting notes from July 18, 2003

This week's plan

  1. We are going to run the experiments again on the pictures that Jenny and Clifford have put together. Yi provided some new structure code that needs to be used for the processing, so that our images are already merged and processed a bit more than before. (inter and intra clustering of regions was used)
  2. Generate webpages so that statistical info as well as the processed images can be viewed together.
  3. Generate symbolic signatures for buildings (where buildings is used to mean office buildings with lots of windows.
The statistical information that we would like from each image/line cluster is the following:
For the image: image size
number of clusters
For each cluster: number of lines
area
density = number of lines / area
relative area = area / image size
Furthermore, we should have a histogram where the orienatations (0-180 degrees) of each line are reflected. We will break the orientations into 8 bins, each 22.5 degrees, which then shows a probability density function. So, instead of saying that a bin containx "n" lines, we would give the computed value n/number of lines in the cluster.

The idea behind all of this is that we want to study the behavior of this classifier and figure out the trade-off between generalization and error.