Re: rectangular ROI

Posted by Michael Schmid on Nov 26, 2012; 11:20am
URL: http://imagej.273.s1.nabble.com/rectangular-ROI-tp3689560p5000958.html

Hi Judith,

when rotating a ROI, there is always a small error due to the finite size of the pixels.
ImageJ uses sub-pixel accuracy for the corners of the rotated rectangle, but the image pixels have integer coordinates, so they are considered either inside or outside the rectangle, even if the border would actually cut a pixel into one portion inside and the rest outside.

Anyhow, this error is much smaller (usually below 1% unless you have very small rectangles) than all the other errors typically involved in image processing. If you want to reduce it, enlarge the image; then the area will be more accurate.

Michael
________________________________________________________________
On Nov 25, 2012, at 23:32, Judith M wrote:

> Hello,
>
> I want to compare the same sized rectangular ROI between different images,
> but some ROI's are rotated and some are not. When I rotate a rectangular ROI
> the area of it changes. Is there any way to rotate a ROI, but keep the area
> the same?
>
> Thanks,
> Judith

--
ImageJ mailing list: http://imagej.nih.gov/ij/list.html