[Insight-users] Image boundary solutions

[Jesse Ross-Jones]

Hi,
So far I have not been able to find a function that does what the
description describes. The closest was PadByRadius from ImageRegion,
however this pads uniformly in any one dimension.

Therefore I started to think about the steps necessary to recreate the
method. This is what I have come up with so far, I would appreciate any
feedback you might have.

0. Original image of the eye (original.im)
1. Threshold image to create binary image (binary.im) (0 =
outside/background, 1 = inside field of view)
2. Dilate binary image by 1 pixel (dilated.im)
3. Subtract binary image from dilated image (dilated.im - binary.im) this
creates a one pixel width boarder around the field of view FOV (boarder.im)
4. For every pixel (i) = 1 in the boarder image find the 8 neighborhood
pixels in both the binary image (binary.im) and the original (original.im)
and mutliply each corresponding neighborhood pixel together. This leaves
only the grey-values of the pixels inside the FOV and removes the
grey-values of the background.
5. Find the average of these multiplied 8 pixels and assign this average to
the current pixel (i),
6. Repeat for remaining pixels inside the boarder

The whole process is then repeated until a large enough boarder is created
around the FOV of the eye.

Can this be done more efficiently or simply?

I look forward to thoughts on this,
much appreciated,
Jesse


On Wed, Jun 5, 2013 at 1:19 PM, Jesse Ross-Jones <jesse.rj at gmail.com> wrote:

> Dear ITK Users and Devs,
>
> I am working on vessel segmentation, using the STARE and DRIVE image
> datasets, and in order to deal with the image boundary I am trying to
> implement the following:
>
> The preprocessing algorithm starts with a region of interest
> (ROI) determined by the camera’s aperture and iteratively grows
> this ROI. Each step of the algorithm consists in the following.
> First, the set of pixels of the exterior border of the ROI is de-
> termined, i.e., pixels that are outside the ROI and are neighbors
> (using four-neighborhood) to pixels inside it. Then, each pixel
> value of this set is replaced with the mean value of its neighbors
> (this time using eight-neighborhood) inside the ROI. Finally, the
> ROI is expanded by inclusion of this altered set of pixels. This
> process is repeated and can be seen as artificially increasing the
> ROI, as shown in Fig. 1(b).
>
> from http://ieeexplore.ieee.org/stamp/stamp.jsp?arnumber=01677727&tag=1
>
> Is there something similar already implemented? I am considering instead
> to use the Wrap Pad Image Filter:
> http://www.itk.org/Doxygen/html/classitk_1_1WrapPadImageFilter.html
> to pad the region around the eye. Would this be a better possibility?
>
> Much Appreciated,
> Jesse
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.itk.org/pipermail/insight-users/attachments/20130607/108e2b7b/attachment.htm>