Thread Subject: image overlay

Steve Eddins posted a nice file (link below) on the FEX
showing how to overlay a binary image on a base image. In
that example, pixels of the base image in RGB format (N x M
x 3) are replaced by the binary image.

In my application, I am zooming in to images such that one
data pixel is displayed on perhaps 2-10 screen pixels (zoom
Z is 2-10). In that mode, a single overlay pixel becomes
also Z screen pixels wide/tall. Does anyone know a way to
generate overlays that uses the full resolution of the
screen even if the underlying data are zoomed in?

One perhaps inelegant idea is to resample the data to have Z
times more points along each edge, and then to overlay as
before with access to each individual pixel. But I suspect
that interpolating each RGB channel would be too time
consuming for my application and see it as a last resort.


Thanks,
J

[imoverlay from Steve Eddins]
http://www.mathworks.com/matlabcentral/fileexchange/loadFile.do?objectId=10502&objectType=file

Thought I would try a second time to see if anyone has ideas
about an image overlay at the screen resolution... thanks.

J

On May 19, 12:57=A0pm, "jay vaughan" <jvaughan5.nos...@gmail.com> wrote:
> Thought I would try a second time to see if anyone has ideas
> about an image overlay at the screen resolution... thanks.
>
> J

---------------------------------------
If you're just plotting an outline, you can use the plot() function
which will plot it as a smooth vector that I don't believe will show
the big quantized blocky pixels when you zoom way in. Can you get the
perimeter of the blob, then scale the numbers, then using poly2mask()
convert it to a bitmap with the higher resolution. This should
produce a smoother boundary because you're basically using the vector-
based boundary to create a higher resolution bitmap.

If you're displaying a pixel-based image, you're out of luck. You
could resample at a higher res but then whatever you did to create the
overlay will still be blocky, unless you do something else like
resample higher and then blur it. That would get you a smoother
overlay but it's no more accurate than the original image in which
pixels got selected for the overlay image, in fact it's less accurate
so don't fool yourself. It's smoother but only because you're making
up intermediate pixels values and locations to get that smoother
boundary. Since you're making up pixels that weren't in your original
image, there's no guarantee that the boundary on your processed image
would correspond better to the image as if you HAD started out with
the higher resolution on the original image than if you just used the
blocky one.
Regards,
ImageAnalyst