An SVP plan 9/29/2001

I have demonstrated that it's not hard to adapt some code from Rodrigues' book to load a jpeg image into a Java2D BufferedImage, as described here. What I now suggest is the following:

• Dan Crispell should work to dig into the other properties of the image such as color versus gray levels. Then he should work, to produce a 2d array of doubles that holds the total intensity values for each pixel. The code from Chap. 8 in ImageDataViewer.java ImageInfoUtil.java should contain enough info to do this. This could be done by defining a class ArrayImage and adapting the Rodrigues code to function as methods of this class. The class methods would be able to read in a jpeg image, find out its properties and eventually produce the 2d array of doubles.



• Bob Futrelle will assume that such an array of doubles exists and begin to design and build pixelbots or "fleas" that can discover vertical and horizontal straight lines and the points at which the straight line model fails (line ends, intersections, projections, etc.). He can test them on real diagrams as soon as Dan can make them available, and in the meantime can test them on synthetic clean and noisy diagrams.

New file space available for work: I have a lot of pretty much unrestricted file space that I've set up in /proj/futrelle/svp/. I've set the group ownership of the directory to bklweb and made it fully accessible to anyone in that group. But you (Dan) do have to execute
newgrp bklweb
to get yourself in that group. Once in it, any files or directories you create will be owned by you but accessible to the group. To check your group memberships, run the groups command. To see who's bklweb, run
ypcat group |grep bklweb

In the svp folder is a folder lhr (for Larry H. Rodrigues) that contains a few files and another folder src with all the sources from the book in it.