help-gift
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [help-GIFT] Searching for similarities (fwd)


From: Andreas Enge
Subject: Re: [help-GIFT] Searching for similarities (fwd)
Date: Thu, 18 Oct 2001 10:21:37 +0200 (CEST)

On Wed, 17 Oct 2001, Wolfgang Mueller wrote:
> > I might give it a try, although I am somewhat sceptical about the
> > outcome. At least it is simple to test...
>
> ...and it does not work (I tested it with resizing to 10x10, and it fails
> miserably, so in the very least, you have to choose the parameters
> carefully). So you won't get around doing some feature extraction. However, I
> doubt that Viper's features are really the best for what you want to do (at
> least in terms of value for processor time).

I carried out a few tests yesterday, and it works in very special cases.
The problem with cryptographic hash functions is that they detect even one
altered bit (this is precisely what they were designed for...), so even
an unnoticeable shift in one pixel will give a completely different result.
It is thus mandatory to reduce the colour palette.

My first (successful) experience went as follows: I took my photo, which
has about 100 times 150 pixels, from my homepage, converted it to jpg and
back to png. This resulted was a coarser picture of twice the file size.
With gimp, I then pixelised both images into blocks of 10 times 10 pixels
and reduced the palette to black and white, so that I was left with a few
black and white blocks. After this, both pictures were identical, and
saving them as gifs resulted in twice the same file.

Second, I slightly altered the saturation of the photo and made it a little
redder. Now, the same procedure failed, because one or two squares switched
from white to black or vice versa.

After all, I agree that comparison for being identical is not a valid
option even when heavily altering the files.

Greetings,

Andreas


_______________________________________________________________________________

Andreas Enge
Laboratoire d'Informatique (LIX)
Ecole Polytechnique
91128 Palaiseau CEDEX
France

Phone:  +33 1 69 33 34 79
Fax:    +33 1 69 33 30 14
E-mail: address@hidden
URL:    http://www.math.uni-augsburg.de/~enge




reply via email to

[Prev in Thread] Current Thread [Next in Thread]