.comment-link {margin-left:.6em;}

tales from the woods

Friday, September 05, 2003

a spitting image

Today, in office, i wanted to measure a certain distance (say 10 cm.)
Since i didn't have a scale handy, i tried googling for the image of a scale that i could print to paper and use like a normal scale(really!). I couldn't find a single image that was in true-scale. Why? I think there's potential in a image-bank site that specializes in true-scale images.

Another thing: If i do publish true-scale images, there is potential for the image to get distorted with scaling. What i mean is, if it's originally a 200x200 image, it can be resized with photoshop or even when displaying it in a html page.
My question is: Is there any way that while creating the original image, we give it say a background pattern from which we could make out that the image is distorted? In other words, can we simply look at any distorted image and clearly say that it's been distorted?

Let's break this down into 3 cases:-
1. An image is either stretched or compressed horizontally only.
2. An image is either stretched or compressed vertically only.
3. An image is either stretched or compressed in equal proportions in both the horizontal and vertical directions. In other words, the image is scaled.

Common sense tells me that the first 2 cases should be much easier to detect than the third. A gut feel also tells me that it IS possible to detect the first 2 cases. For example, a face clearly looks distorted when stretched horizontally or vertically. But then, a tree may not. So, what is it in a face that enables us to determine reasonable mutations in it's proportion? Is it symmetry? Or is it our expertise in human faces (we stare at them for pretty much our entire lives)? Or is it something else like an innate sense of proportion itself?

This problem now defines itself as a problem in understanding human congnitive ability. Since i'm a layman in this field, i can only use my common sense to try and solve this problem. A simplistic solution that occurs to me is using a pattern of 45 degree angled lines (say diamonds) as the image background. If the image is distorted horizontally, the angle would change. Since we're reasonably accurate at judging a 45 degree angle, we should be able to determine if the image has been stretched. We could even use circles instead of diamonds. Does this solution make sense?

There should be more and better ways to solve this problem, but it's not coming to me right now. Let me know if you can think of something too.

0 Comments:

Post a Comment

Links to this post:

Create a Link

<< Home