Comments on: Unplanned planning

By: greggT

greggT — Wed, 30 Dec 2009 02:21:46 +0000

In medical imaging, we talk a point in either of 2 interpretations: a single point or a 1×1 area whose average is represented at a single location.

The former requires that you specify where in the square that measurement is taken. It can only be represented as a float, but you can, if you have rules, store them in int’s such that the inverse operation is valid.

The latter is because medical images represent actual body tissues and usually average a measurement of these tissues (color, attenuation, opacity, etc) over an area (called an aperture), not at a single point. Why? Because ALL measurements (not just in images) are taken over an area (or volume or time span or…). Think of the aperture ring on a camera — does it have a zero setting? No. Even a pinhole camera has an aperture of the size of a pin, which is NOT zero. A zero sized aperture is not a hole, and so nothing can pass thru it.

So you cannot make a rectangle of the latter concept, and must be careful about the former.

By: Michael

Michael — Sat, 12 Dec 2009 16:41:57 +0000

Anonymous has it right–just use x/y and height/width.

By: Les

Les — Tue, 08 Dec 2009 12:30:35 +0000

I thought the lower right corner would be x=10,y=0.

And the number of points in the image is determined by how your language declares an array[10,10]. Zero based? Is the 10 inclusive or not?

By: Trevor

Trevor — Tue, 08 Dec 2009 03:55:24 +0000

“Hopefully, your answers disagreed (or at least you can see how they could disagree).”

I don’t see how they can disagree.

By: Anonymous

Anonymous — Mon, 07 Dec 2009 19:34:34 +0000

And this is why I use struct rect { float left, top, width, height; }