PyQt QImage cna't copy, crash without error code

JonB

@darrenleeleelee1 said in PyQt QImage cna't copy, crash without error code:

@JonB It won't crash!But when I change to
qimage.copy(0, 0, qimage.width(), qimage.height())

then it crash.Even I give to other variable first it still not work
wt = qimage.width()
ht = qimage.height()
qimage = qimage.copy(0, 0, wt, ht)

This all implies that the source qimage you start with is the issue.

Start by looking at exactly what your arr = ds.pixel_array holds/returns. The QImage() constructor you use states "it must remain valid". That looks to me like the danger area. I think when you copy a portion of the image Qt takes a copy of its own. But probably not when you copy the whole thing. For example, are you able to do a Python copy() on it so you have your own separate copy, which you then use for the image? And btw, if you have any threading involved anywhere, please say so......

darrenleeleelee1

@JonB Sorry I am not really get what did you mean.And I think I don't have any threading.(I think..)

JonB

@darrenleeleelee1
I have no idea what is behind your

        ds = dcmread('./5F329172_20170623_CR_2_1_1')
        arr = ds.pixel_array

It's some external library stuff. (And btw I have no idea whether arr.shape[1], arr.shape[0] is the right thing to use. Don't bother to explain, just make sure it's correct for whatever this ds stuff is.)

Try this:

import copy
...
        arr = copy.copy(ds.pixel_array)
# or even:
#        arr = copy.deepcopy(ds.pixel_array)

before you proceed to qimage = QtGui.QImage(arr, ...). Any better, crash-wise?

darrenleeleelee1

@JonB
I add deepcopy,and it also crash. I add some arr attribute for you.

KroMignon

@darrenleeleelee1 I think you should change your code to:

qimage = QtGui.QImage(arr.data, arr.shape[1], arr.shape[0], QtGui.QImage.Format_Grayscale16)

darrenleeleelee1

@KroMignon I change but still crash...What is the different between arr and arr.data?

KroMignon

@darrenleeleelee1 said in PyQt QImage cna't copy, crash without error code:

What is the different between arr and arr.data?

arr is an instance of numpy.ndarray type, according to your code!
And arr.data is the data buffer according to numpy documentation ==> https://numpy.org/doc/stable/reference/generated/numpy.ndarray.html

I think it would be great if you first explain what you want to do.
Perhaps you should take a look at this python script which converts numpy.ndatarray to QImage and vice versa:
https://kogs-www.informatik.uni-hamburg.de/~meine/software/vigraqt/qimage2ndarray.py

darrenleeleelee1

@KroMignon I saw this before, but I don't know how to apply to uint16.

KroMignon

@darrenleeleelee1 said in PyQt QImage cna't copy, crash without error code:

I saw this before, but I don't know how to apply to uint16.

Which version of Qt-Libs are you using?
Format_Grayscale16 is only support starting from Qt 5.13 (cf. QImage.Format)

Because of this note from documentation:

Constructs an image with the given width , height and format , that uses an existing memory buffer, data . The width and height must be specified in pixels, data must be 32-bit aligned, and each scanline of data in the image must also be 32-bit aligned.

Perhaps you should specify the number of bytes per line, like this:

qimage = QtGui.QImage(arr.data, arr.shape[1], arr.shape[0], arr.shape[1]*2, QtGui.QImage.Format_Grayscale16)

darrenleeleelee1

@KroMignon said in PyQt QImage cna't copy, crash without error code:

arr.shape[1]*2

sorry, I read through the QImage API, but I don;t get why I need to add this.
After I add this it seems can work.

KroMignon

@darrenleeleelee1 said in PyQt QImage cna't copy, crash without error code:

sorry, I read through the QImage API, but I don;t get why I need to add this.

Then you don't read carefully ;)
The point is when you only specify width and height: The width and height must be specified in pixels, data must be 32-bit aligned, and each scanline of data in the image must also be 32-bit aligned.

To avoid this, you can use width, height and scanline length in bytes, which is pixel_per_line x pixel_size:

pixel_per_line is arr.shape[1]
pixel_size is uint16 which are 2 bytes

==> scanline_size = arr.shape[1]*2

After I add this it seems can work.

Great

darrenleeleelee1

@KroMignon @J-Hilk @JonB @jsulm thanks all u guys!