[vtkusers] Resampling an image to compare with other one

[Paul Arner]

From: "Paul Arner" <paularner at msn.com>
Sent: Saturday, September 26, 2009 9:56 PM
To: "David Gobbi" <david.gobbi at gmail.com>
Subject: Re: [vtkusers] Resampling an image to compare with other one

> Hi RaC-
>
> If I understand this correctly,  you wanted to take image B (W = 120, H = 
> 80) to image A: (W = 80, H = 120),
>
> so the x-(width) scale should be (119/79) and the y-(height) scale should 
> be (79/119), as David argues below.
>
> From the description, you were using (120/80) = 3/2 and (80/120) = 2/3.
>
> (B max column value is 119), so    119 * 2/3 = 79.333 ~ 79, so the width 
> is correct, but
>
> (B max row value is 79), so 79 * 3/2 = 118.5 ~ 118, not 119 so the height 
> is not correct, as you observed.
>
>
>
> Paul
>
> --------------------------------------------------
> From: "David Gobbi" <david.gobbi at gmail.com>
> Sent: Saturday, September 26, 2009 8:57 PM
> To: <ra.corredor at gmail.com>
> Cc: <vtkusers at vtk.org>
> Subject: Re: [vtkusers] Resampling an image to compare with other one
>
>> Hi RaC,
>>
>> This has to do with the peculiar mathematics of image resizing.  To
>> simplify things, I'll only discuss the image widths, since the height
>> is treated in exactly the same way.
>>
>> Image A has a columns in the range [0,79], and image B has columns
>> [0,119].  In order to stretch image A to the size of image B, you need
>> to set a magnification factor that will stretch the rightmost column
>> of A out to the rightmost column of B.  The correct factor to apply is
>> 119/79 which is 1.506329.  If you feel doubtful about this, then just
>> try it.  Note that the leftmost column stays in the same place, as it
>> should, because 0*(119/79) = 0.
>>
>> So if you apply a magnification factor of 2, then [0,79]*2 = [0,158]
>> which is why vtkImageResample will give you and image that is 159
>> pixels across.
>>
>> To make a long story short: when you are resizing images or doing any
>> other sort of geometrical changes to them, you have to measure the
>> width of the image as the distance between the leftmost pixel and the
>> rightmost pixel, which is equal to the number of pixels subtract one.
>>
>> Basically, VTK does not include the "borders" of the images in its
>> data coordinate system.  In this way it is different from other
>> coordinate systems that do include the image border, like texture
>> coordinates.  The way that VTK does coordinates is the same as medical
>> images, e.g. like CT or MRI, while the other way is commonly used in
>> computer graphics and digital photography.
>>
>>  David
>>
>>
>> On Sat, Sep 26, 2009 at 4:27 PM,  <ra.corredor at gmail.com> wrote:
>>> Hi
>>>
>>> I have two images of different sizes. Image A:80x120 Image B:120x80. I 
>>> must
>>> compare both in the same size, so I have to resample or resize one of 
>>> the
>>> images with the size values of the other one. When I use the
>>> vtkImageResample with B, the width changes to 80, but the height to 119 
>>> !!
>>> ... And when I use different factor values in the method
>>> SetAxisMagnificationFactor, I note these results:
>>> Factor Obtained Expected Height Value
>>> 0,5 40 40
>>> 1 80 80
>>> 1,5 119 120
>>> 2 159 160
>>> 2,01 159 161
>>> 2,1 166 168
>>> 2,7 214 216
>>> 3 238 240
>>> 3,5 277 280
>>> 4 317 320
>>> 4,1 324 328
>>> So, when the factor increases, the number of pixels obtained with the
>>> SetAxisMagnificationFactor is less than the expected value.
>>>
>>> Do you know how can I resize the B image keeping the extent of the A 
>>> image
>>> and not losing some rows or columns in the process??
>>>
>>>
>>> Thanks!!
>>>
>>> RaC
>>> _______________________________________________
>>> Powered by www.kitware.com
>>>
>>> Visit other Kitware open-source projects at
>>> http://www.kitware.com/opensource/opensource.html
>>>
>>> Please keep messages on-topic and check the VTK FAQ at:
>>> http://www.vtk.org/Wiki/VTK_FAQ
>>>
>>> Follow this link to subscribe/unsubscribe:
>>> http://www.vtk.org/mailman/listinfo/vtkusers
>>>
>>>
>> _______________________________________________
>> Powered by www.kitware.com
>>
>> Visit other Kitware open-source projects at 
>> http://www.kitware.com/opensource/opensource.html
>>
>> Please keep messages on-topic and check the VTK FAQ at: 
>> http://www.vtk.org/Wiki/VTK_FAQ
>>
>> Follow this link to subscribe/unsubscribe:
>> http://www.vtk.org/mailman/listinfo/vtkusers
>>