[Insight-developers] Region bug in PNGImageIO

[Joshua Cates]

I vote for consistency with the rest of the pipeline: writing just the
RequestedRegion.  Even if the buffered region contains extra and valid
data, the RequestedRegion is by definition the only region of interest to
the user.  The image copy would only be needed in cases where
BufferedRegion was larger than RequestedRegion.

Josh.

______________________________
 Josh Cates			
 School of Computer Science	
 University of Utah
 Email: cates@sci.utah.edu
 Phone: (801) 587-7697
 URL:   http://www.sci.utah.edu/~cates


On Mon, 31 Mar 2003, Miller, James V (Research) wrote:

> Here is what I understand of the ImageIO architecture:
> 
> There is a basic disconnect between the ImageIO's view of an ITK image
> and the pipeline's view of an ITK image.
> 
> The interface between the ImageFileWriter and the ImageIO is at the
> level of a "buffer".  The ImageFileWriter sets up an IORegion, spacing,
> and origin and passes them to the ImageIO object.  Then the ImageFileWriter
> passes down a pointer to a buffer to write. 
> 
> So what region and pointer should be set on the ImageIO object? The simple 
> answers are the LargestPossibleRegion or the RequestedRegion. But the
> pipeline
> does not guarentee that the input to the writer will be exactly the
> RequestedRegion.
> The upstream filter has the option to produce more than the RequestedRegion.
> 
> Furthermore, the RequestedRegion may be smaller than the
> LargestPossibleRegion.
> 
> So we have a choice of writing out the BufferedRegion or the
> RequestedRegion.
> If we choose the former, then we can just pass the normal buffer pointer 
> down to the ImageIO object and set the IORegion to the BufferedRegion. 
> It is actually not quite as simple as this because for most of the ImageIO
> writers, the IORegion needs to start at index zero. So we would need
> to adjust the IORegion. If we choose the latter, then we need to do one of 
> two things:
> 
> 1) Create a new buffer that is contiguous in memory and only contains pixels
> of the RequestedRegion. Then set the IORegion such that its indices start 
> at zero.
> 
> 2) Pass two regions down to the ImageIO objects.  One that describes the 
> buffer and one that describes what subregion you want to write. Thus, 
> the ImageIO objects would need to understand the difference between 
> a BufferedRegion and a RequestedRegion.
> 
> Another question that arises in considering the output of the
> RequestedRegion:
> If the RequestedRegion is a subregion of a larger image, should the output
> image
> be the size of the larger image with only the pixels in the subregion set?
> (i.e.
> padded with zeros?).
> 
> 
> 
> 
> 
> 
> > -----Original Message-----
> > From: Joshua Cates [mailto:cates@sci.utah.edu]
> > Sent: Friday, March 28, 2003 11:13 PM
> > To: Insight-Developers
> > Subject: [Insight-developers] Region bug in PNGImageIO
> > 
> > 
> > Hi,
> > 
> > As discussed in todays TCON, PNGImageIO does not produce a 
> > valid output 
> > when given an image whose start index is not 0,0.  
> > Insight/Examples/ImageRegionIterator illustrates this problem.
> > 
> > Josh.
> > 
> > ______________________________
> >  Josh Cates			
> >  School of Computer Science	
> >  University of Utah
> >  Email: cates@sci.utah.edu
> >  Phone: (801) 587-7697
> >  URL:   http://www.sci.utah.edu/~cates
> > 
> > 
> > 
> > _______________________________________________
> > Insight-developers mailing list
> > Insight-developers@public.kitware.com
> > http://public.kitware.com/mailman/listinfo/insight-developers
> > 
> _______________________________________________
> Insight-developers mailing list
> Insight-developers@public.kitware.com
> http://public.kitware.com/mailman/listinfo/insight-developers
>