Image recognition tool

[Dar Scott]

On Saturday, July 19, 2003, at 10:47 PM, Vikram Singh wrote:

> It will also allow you (after formatting) to maintain a list of human 
> readable
> chunks of characters without necessarily maintaining the corresponding 
> image
> in that environment (eg that image may be in one of many substacks 
> which are
> not open). You can then compare the correspondingly formatted 
> md5digest of
> incoming images with the list to know if a copy exists. I think 
> searching
> through a few thousand lines will take less than a few seconds! Isnt 
> rev
> powerful?

This is a very good idea.  To find a duplicate among n images will take 
n*(n-1)/2 comparisons.  Comparing the md5 will be much faster.  This 
will work either on binary files or on imageData() & alphaData() & 
formatedWidth().

You might even speed that up by sorting the md5 and looking that up or 
by using arrays.

Dar Scott