MantisBT - ITK
View Issue Details
0010393ITKpublic2010-03-09 08:032010-11-20 12:21
Jef 
Bill Lorensen 
normalmajoralways
assignedopen 
ITK-3-16 
 
0010393: itkDownsampleImageFilter is producing an output incorrect along Y-axis
The filter produces an image in which the original image content is not fully included along the Y-axis. Instead the result seems stretched along that dimension, and only half is portraited (see screenshot).

The change in behavior happened somewhere between ITK 3.16 and ITK 3.8
Information concerning the attached screenshot

As the names of the images (visible in the bottom left corner of the images) suggest, the top left picture is the input image, top right the output image when using the 3.8 version of the DownsampleImageFilter, bottom left the output image when using the 3.16 version of the DownsampleImageFilter and bottom right the output image when using the 3.16 version of the UpsampleImageFilter.

The origin remains unchanged for all images (even with the centered coordinate system). The size and spacing of the images are, respectively:

209x155x169 2x2x2 (input)
104x77x84 4x4x4 (output_downsample_3_8)
104x77x84 4x4x4 (output_downsample_3_16)
418x310x338 1x1x1 (output_upsample_3_16)

It seems that the images have the correct size and spacing. However based on their appearance, the "image content" is incomplete in the case of the 3.16 version. Only half of the image is shown in the Y-dimension. This is not the case for the 3.8 version or for the upsampler. I am unable to say if these images are correct however.
No tags attached.
? Screenshot-vv_DownsampleImageFilter (203,833) 2010-03-09 08:03
https://public.kitware.com/Bug/file/2941/Screenshot-vv_DownsampleImageFilter
Issue History
2010-03-09 08:03JefNew Issue
2010-03-09 08:03JefFile Added: Screenshot-vv_DownsampleImageFilter
2010-11-02 14:05Hans JohnsonStatusnew => assigned
2010-11-02 14:05Hans JohnsonAssigned To => Hans Johnson
2010-11-20 12:20Hans JohnsonAssigned ToHans Johnson => Bill Lorensen
2010-11-20 12:21Hans JohnsonNote Added: 0023464

Notes
(0023464)
Hans Johnson   
2010-11-20 12:21   
Bill,

Would you mind investigating this bug to determine if it is still relevant?

Thanks,
Hans