[pyqtgraph] Display Images from Disk, either list of files or large binary file

[Dennis Norton]

I'm wondering if there is a convenient way to use the ImageView class to
display images either from a list of files in a directory or from a large
binary file consisting of multiple frames.

Could we create a generator which yields the numpy arrays to display?

Thanks.

[Patrick]

Hi,

You can load images into numpy arrays (usable by ImageItem, ImageView etc)
using the scipy.ndimage.imread function. I think it relies on having the
Python Image Library (PIL) installed. A more modern version of that library
is Pillow, which you could also use to load image data.

Limitations are though that the ImageView expects each image to have the
same dimensions.

Simple modification of ImageView example to demonstrate (specify your image
files on command line).

#!/usr/bin/env python3


import numpy as np
from pyqtgraph.Qt import QtCore, QtGui
import pyqtgraph as pg
from argparse import ArgumentParser
from scipy.ndimage import imread

argparser = ArgumentParser(description='Load images and display using
pyqtgraph ImageView widget.')
argparser.add_argument('input_images', help='Input filename(s) of images',
nargs='+')
args = argparser.parse_args()

imagelist = []
for filename in args.input_images:
print("Loading {}...".format(filename))
image = imread(filename)
imagelist.append(image)
print('Image shape: {}{}'.format(imagelist[len(imagelist)-1].shape, ''
if imagelist[len(imagelist)-1].shape == imagelist[0].shape else ' *'))
data=np.stack([ i for i in imagelist if i.shape==imagelist[0].shape ], 0)
if len(args.input_images) != data.shape[0]:
print('Warning: Shape of all input files must match. Those marked with
* are skipped.')

# Interpret image data as row-major instead of col-major
pg.setConfigOptions(imageAxisOrder='row-major')

app = QtGui.QApplication([])

## Create window with ImageView widget
win = QtGui.QMainWindow()
win.resize(800,800)
imv = pg.ImageView()
win.setCentralWidget(imv)
win.show()
win.setWindowTitle('pyqtgraph example: ImageView')

## Display the data
imv.setImage(data)

## Set a custom color map
colors = [
(0, 0, 0),
(45, 5, 61),
(84, 42, 55),
(150, 87, 60),
(208, 171, 141),
(255, 255, 255)
]
cmap = pg.ColorMap(pos=np.linspace(0.0, 1.0, 6), color=colors)
imv.setColorMap(cmap)

## Start Qt event loop unless running in interactive mode.
if __name__ == '__main__':
import sys
if (sys.flags.interactive != 1) or not hasattr(QtCore, 'PYQT_VERSION'):
QtGui.QApplication.instance().exec_()

Post by Dennis Norton
I'm wondering if there is a convenient way to use the ImageView class to
display images either from a list of files in a directory or from a large
binary file consisting of multiple frames.
Could we create a generator which yields the numpy arrays to display?
Thanks.

[Dennis Norton]

Thanks, Patrick.

I should have been more explicit about what I was asking. The method you
outlined I was aware of.

What I'm asking has to do if you have such a large number of files or a
single file that is so large it can't be (or is impractical to be) read
into memory.

The idea I tried, and works thus far is to use a numpy memmap object.

So imagine a file "VeryLargeImageFile.img", which consists of the binary
data of type unsigned integer (uint16). Let's say this was captured with a
camera of format frame shape (512,640), and there are 10,000 frames stored
in the file.

fp = np.memmap('VeryLargeImageFile.img', dtype='unt16', mode='r',
offset=0, shape=(10000,512,640))

The variable can be passed to ImageView item by imv.setImage(fp).

However, for 32-bit operating systems, fp would still be limited to 2Gb.

So my real question is if there is a method to pass a generator which reads
a frame at a time to the ImageView or, if I had a large list of binary
files, could I create a generator which will cycle through reading each
file and return a numpy array of the data read for each file.

Is what I'm asking making more sense, now? If not, let me know and I'll
try to clarify by uploading my current code.

Thanks again.

-Dennis

Post by Patrick
Hi,
You can load images into numpy arrays (usable by ImageItem, ImageView etc)
using the scipy.ndimage.imread function. I think it relies on having the
Python Image Library (PIL) installed. A more modern version of that library
is Pillow, which you could also use to load image data.
Limitations are though that the ImageView expects each image to have the
same dimensions.
Simple modification of ImageView example to demonstrate (specify your
image files on command line).
#!/usr/bin/env python3
import numpy as np
from pyqtgraph.Qt import QtCore, QtGui
import pyqtgraph as pg
from argparse import ArgumentParser
from scipy.ndimage import imread
argparser = ArgumentParser(description='Load images and display using
pyqtgraph ImageView widget.')
argparser.add_argument('input_images', help='Input filename(s) of images',
nargs='+')
args = argparser.parse_args()
imagelist = []
print("Loading {}...".format(filename))
image = imread(filename)
imagelist.append(image)
print('Image shape: {}{}'.format(imagelist[len(imagelist)-1].shape, ''
if imagelist[len(imagelist)-1].shape == imagelist[0].shape else ' *'))
data=np.stack([ i for i in imagelist if i.shape==imagelist[0].shape ], 0)
print('Warning: Shape of all input files must match. Those marked
with * are skipped.')
# Interpret image data as row-major instead of col-major
pg.setConfigOptions(imageAxisOrder='row-major')
app = QtGui.QApplication([])
## Create window with ImageView widget
win = QtGui.QMainWindow()
win.resize(800,800)
imv = pg.ImageView()
win.setCentralWidget(imv)
win.show()
win.setWindowTitle('pyqtgraph example: ImageView')
## Display the data
imv.setImage(data)
## Set a custom color map
colors = [
(0, 0, 0),
(45, 5, 61),
(84, 42, 55),
(150, 87, 60),
(208, 171, 141),
(255, 255, 255)
]
cmap = pg.ColorMap(pos=np.linspace(0.0, 1.0, 6), color=colors)
imv.setColorMap(cmap)
## Start Qt event loop unless running in interactive mode.
import sys
if (sys.flags.interactive != 1) or not hasattr(QtCore, 'PYQT_VERSION'
QtGui.QApplication.instance().exec_()

Post by Dennis Norton
I'm wondering if there is a convenient way to use the ImageView class to
display images either from a list of files in a directory or from a large
binary file consisting of multiple frames.
Could we create a generator which yields the numpy arrays to display?
Thanks.
--

You received this message because you are subscribed to a topic in the
Google Groups "pyqtgraph" group.
To unsubscribe from this topic, visit
https://groups.google.com/d/topic/pyqtgraph/1PuRof6NA5w/unsubscribe.
To unsubscribe from this group and all its topics, send an email to
To view this discussion on the web visit
https://groups.google.com/d/msgid/pyqtgraph/d1681786-8d9c-4a97-a3b5-78f32223e481%40googlegroups.com
<https://groups.google.com/d/msgid/pyqtgraph/d1681786-8d9c-4a97-a3b5-78f32223e481%40googlegroups.com?utm_medium=email&utm_source=footer>
.
For more options, visit https://groups.google.com/d/optout.

[Patrick]

Hi Dennis,

Yes, that makes sense now... and unfortunately I can't see an easy way to
plumb that into the ImageView widget. An idea though, if you look at the
code for ImageView
(http://www.pyqtgraph.org/documentation/_modules/pyqtgraph/imageview/ImageView.html#ImageView.setImage)
the image provided does not strictly need to be an numpy array (just needs
certain attributes), so perhaps you could make a shim class which
extends/mimics a numpy ndarray, catching calls to these attributes and
mapping to the appropriate disk image file.
Of course, maybe someone has already made something like this...
http://docs.dask.org/en/latest/array.html looks interesting, though I've
never used it.

Good luck!
Patrick

Post by Dennis Norton
Thanks, Patrick.
I should have been more explicit about what I was asking. The method you
outlined I was aware of.
What I'm asking has to do if you have such a large number of files or a
single file that is so large it can't be (or is impractical to be) read
into memory.
The idea I tried, and works thus far is to use a numpy memmap object.
So imagine a file "VeryLargeImageFile.img", which consists of the binary
data of type unsigned integer (uint16). Let's say this was captured with a
camera of format frame shape (512,640), and there are 10,000 frames stored
in the file.
fp = np.memmap('VeryLargeImageFile.img', dtype='unt16', mode='r',
offset=0, shape=(10000,512,640))
The variable can be passed to ImageView item by imv.setImage(fp).
However, for 32-bit operating systems, fp would still be limited to 2Gb.
So my real question is if there is a method to pass a generator which
reads a frame at a time to the ImageView or, if I had a large list of
binary files, could I create a generator which will cycle through reading
each file and return a numpy array of the data read for each file.
Is what I'm asking making more sense, now? If not, let me know and I'll
try to clarify by uploading my current code.
Thanks again.
-Dennis

Post by Patrick
Hi,
You can load images into numpy arrays (usable by ImageItem, ImageView
etc) using the scipy.ndimage.imread function. I think it relies on having
the Python Image Library (PIL) installed. A more modern version of that
library is Pillow, which you could also use to load image data.
Limitations are though that the ImageView expects each image to have the
same dimensions.
Simple modification of ImageView example to demonstrate (specify your
image files on command line).
#!/usr/bin/env python3
import numpy as np
from pyqtgraph.Qt import QtCore, QtGui
import pyqtgraph as pg
from argparse import ArgumentParser
from scipy.ndimage import imread
argparser = ArgumentParser(description='Load images and display using
pyqtgraph ImageView widget.')
argparser.add_argument('input_images', help='Input filename(s) of images'
, nargs='+')
args = argparser.parse_args()
imagelist = []
print("Loading {}...".format(filename))
image = imread(filename)
imagelist.append(image)
print('Image shape: {}{}'.format(imagelist[len(imagelist)-1].shape,
'' if imagelist[len(imagelist)-1].shape == imagelist[0].shape else ' *'))
data=np.stack([ i for i in imagelist if i.shape==imagelist[0].shape ], 0)
print('Warning: Shape of all input files must match. Those marked
with * are skipped.')
# Interpret image data as row-major instead of col-major
pg.setConfigOptions(imageAxisOrder='row-major')
app = QtGui.QApplication([])
## Create window with ImageView widget
win = QtGui.QMainWindow()
win.resize(800,800)
imv = pg.ImageView()
win.setCentralWidget(imv)
win.show()
win.setWindowTitle('pyqtgraph example: ImageView')
## Display the data
imv.setImage(data)
## Set a custom color map
colors = [
(0, 0, 0),
(45, 5, 61),
(84, 42, 55),
(150, 87, 60),
(208, 171, 141),
(255, 255, 255)
]
cmap = pg.ColorMap(pos=np.linspace(0.0, 1.0, 6), color=colors)
imv.setColorMap(cmap)
## Start Qt event loop unless running in interactive mode.
import sys
if (sys.flags.interactive != 1) or not hasattr(QtCore, 'PYQT_VERSION'
QtGui.QApplication.instance().exec_()

Post by Dennis Norton
I'm wondering if there is a convenient way to use the ImageView class to
display images either from a list of files in a directory or from a large
binary file consisting of multiple frames.
Could we create a generator which yields the numpy arrays to display?
Thanks.
--

You received this message because you are subscribed to a topic in the
Google Groups "pyqtgraph" group.
To unsubscribe from this topic, visit
https://groups.google.com/d/topic/pyqtgraph/1PuRof6NA5w/unsubscribe.
To unsubscribe from this group and all its topics, send an email to
To view this discussion on the web visit
https://groups.google.com/d/msgid/pyqtgraph/d1681786-8d9c-4a97-a3b5-78f32223e481%40googlegroups.com
<https://groups.google.com/d/msgid/pyqtgraph/d1681786-8d9c-4a97-a3b5-78f32223e481%40googlegroups.com?utm_medium=email&utm_source=footer>
.
For more options, visit https://groups.google.com/d/optout.

--
-Dennis

[Dennis Norton]

Thanks again for your response.

I did think about creating a custom class which would have the attributes
ImageView is looking for, or just create a custom ImageView-esque class
which would handle a generator.

I'll look into the array link you sent.

Best regards.

Post by Patrick
Hi Dennis,
Yes, that makes sense now... and unfortunately I can't see an easy way to
plumb that into the ImageView widget. An idea though, if you look at the
code for ImageView (
http://www.pyqtgraph.org/documentation/_modules/pyqtgraph/imageview/ImageView.html#ImageView.setImage)
the image provided does not strictly need to be an numpy array (just needs
certain attributes), so perhaps you could make a shim class which
extends/mimics a numpy ndarray, catching calls to these attributes and
mapping to the appropriate disk image file.
Of course, maybe someone has already made something like this...
http://docs.dask.org/en/latest/array.html looks interesting, though I've
never used it.
Good luck!
Patrick

Post by Dennis Norton
Thanks, Patrick.
I should have been more explicit about what I was asking. The method you
outlined I was aware of.
What I'm asking has to do if you have such a large number of files or a
single file that is so large it can't be (or is impractical to be) read
into memory.
The idea I tried, and works thus far is to use a numpy memmap object.
So imagine a file "VeryLargeImageFile.img", which consists of the binary
data of type unsigned integer (uint16). Let's say this was captured with a
camera of format frame shape (512,640), and there are 10,000 frames stored
in the file.
fp = np.memmap('VeryLargeImageFile.img', dtype='unt16', mode='r',
offset=0, shape=(10000,512,640))
The variable can be passed to ImageView item by imv.setImage(fp).
However, for 32-bit operating systems, fp would still be limited to 2Gb.
So my real question is if there is a method to pass a generator which
reads a frame at a time to the ImageView or, if I had a large list of
binary files, could I create a generator which will cycle through reading
each file and return a numpy array of the data read for each file.
Is what I'm asking making more sense, now? If not, let me know and I'll
try to clarify by uploading my current code.
Thanks again.
-Dennis

Post by Patrick
Hi,
You can load images into numpy arrays (usable by ImageItem, ImageView
etc) using the scipy.ndimage.imread function. I think it relies on having
the Python Image Library (PIL) installed. A more modern version of that
library is Pillow, which you could also use to load image data.
Limitations are though that the ImageView expects each image to have the
same dimensions.
Simple modification of ImageView example to demonstrate (specify your
image files on command line).
#!/usr/bin/env python3
import numpy as np
from pyqtgraph.Qt import QtCore, QtGui
import pyqtgraph as pg
from argparse import ArgumentParser
from scipy.ndimage import imread
argparser = ArgumentParser(description='Load images and display using
pyqtgraph ImageView widget.')
argparser.add_argument('input_images', help='Input filename(s) of
images', nargs='+')
args = argparser.parse_args()
imagelist = []
print("Loading {}...".format(filename))
image = imread(filename)
imagelist.append(image)
print('Image shape: {}{}'.format(imagelist[len(imagelist)-1].shape,
'' if imagelist[len(imagelist)-1].shape == imagelist[0].shape else ' *'
))
data=np.stack([ i for i in imagelist if i.shape==imagelist[0].shape ], 0
)
print('Warning: Shape of all input files must match. Those marked
with * are skipped.')
# Interpret image data as row-major instead of col-major
pg.setConfigOptions(imageAxisOrder='row-major')
app = QtGui.QApplication([])
## Create window with ImageView widget
win = QtGui.QMainWindow()
win.resize(800,800)
imv = pg.ImageView()
win.setCentralWidget(imv)
win.show()
win.setWindowTitle('pyqtgraph example: ImageView')
## Display the data
imv.setImage(data)
## Set a custom color map
colors = [
(0, 0, 0),
(45, 5, 61),
(84, 42, 55),
(150, 87, 60),
(208, 171, 141),
(255, 255, 255)
]
cmap = pg.ColorMap(pos=np.linspace(0.0, 1.0, 6), color=colors)
imv.setColorMap(cmap)
## Start Qt event loop unless running in interactive mode.
import sys
if (sys.flags.interactive != 1) or not hasattr(QtCore,
QtGui.QApplication.instance().exec_()

Post by Dennis Norton
I'm wondering if there is a convenient way to use the ImageView class
to display images either from a list of files in a directory or from a
large binary file consisting of multiple frames.
Could we create a generator which yields the numpy arrays to display?
Thanks.
--

You received this message because you are subscribed to a topic in the
Google Groups "pyqtgraph" group.
To unsubscribe from this topic, visit
https://groups.google.com/d/topic/pyqtgraph/1PuRof6NA5w/unsubscribe.
To unsubscribe from this group and all its topics, send an email to
To view this discussion on the web visit
https://groups.google.com/d/msgid/pyqtgraph/d1681786-8d9c-4a97-a3b5-78f32223e481%40googlegroups.com
<https://groups.google.com/d/msgid/pyqtgraph/d1681786-8d9c-4a97-a3b5-78f32223e481%40googlegroups.com?utm_medium=email&utm_source=footer>
.
For more options, visit https://groups.google.com/d/optout.

--
-Dennis
--

You received this message because you are subscribed to a topic in the
Google Groups "pyqtgraph" group.
To unsubscribe from this topic, visit
https://groups.google.com/d/topic/pyqtgraph/1PuRof6NA5w/unsubscribe.
To unsubscribe from this group and all its topics, send an email to
To view this discussion on the web visit
https://groups.google.com/d/msgid/pyqtgraph/4a8bb184-c59f-4bbc-ae0d-2c8138fc0f9b%40googlegroups.com
<https://groups.google.com/d/msgid/pyqtgraph/4a8bb184-c59f-4bbc-ae0d-2c8138fc0f9b%40googlegroups.com?utm_medium=email&utm_source=footer>
.
For more options, visit https://groups.google.com/d/optout.

[Aaron Adamson]

Another option - you can possibly embed your data in an hdf5 file, and then
read / downsample the hdf5 file on demand based on your range settings. You
can look at the "HDF5 big data" example in pyqtgraph.examples to see it
done with a line plot. I am working on a similar problem, and will post an
example if I get a good solution.

Post by Dennis Norton
Thanks again for your response.
I did think about creating a custom class which would have the attributes
ImageView is looking for, or just create a custom ImageView-esque class
which would handle a generator.
I'll look into the array link you sent.
Best regards.

Post by Patrick
Hi Dennis,
Yes, that makes sense now... and unfortunately I can't see an easy way to
plumb that into the ImageView widget. An idea though, if you look at the
code for ImageView (
http://www.pyqtgraph.org/documentation/_modules/pyqtgraph/imageview/ImageView.html#ImageView.setImage)
the image provided does not strictly need to be an numpy array (just needs
certain attributes), so perhaps you could make a shim class which
extends/mimics a numpy ndarray, catching calls to these attributes and
mapping to the appropriate disk image file.
Of course, maybe someone has already made something like this...
http://docs.dask.org/en/latest/array.html looks interesting, though I've
never used it.
Good luck!
Patrick

Post by Dennis Norton
Thanks, Patrick.
I should have been more explicit about what I was asking. The method
you outlined I was aware of.
What I'm asking has to do if you have such a large number of files or a
single file that is so large it can't be (or is impractical to be) read
into memory.
The idea I tried, and works thus far is to use a numpy memmap object.
So imagine a file "VeryLargeImageFile.img", which consists of the binary
data of type unsigned integer (uint16). Let's say this was captured with a
camera of format frame shape (512,640), and there are 10,000 frames stored
in the file.
fp = np.memmap('VeryLargeImageFile.img', dtype='unt16', mode='r',
offset=0, shape=(10000,512,640))
The variable can be passed to ImageView item by imv.setImage(fp).
However, for 32-bit operating systems, fp would still be limited to 2Gb.
So my real question is if there is a method to pass a generator which
reads a frame at a time to the ImageView or, if I had a large list of
binary files, could I create a generator which will cycle through reading
each file and return a numpy array of the data read for each file.
Is what I'm asking making more sense, now? If not, let me know and I'll
try to clarify by uploading my current code.
Thanks again.
-Dennis

Post by Patrick
Hi,
You can load images into numpy arrays (usable by ImageItem, ImageView
etc) using the scipy.ndimage.imread function. I think it relies on having
the Python Image Library (PIL) installed. A more modern version of that
library is Pillow, which you could also use to load image data.
Limitations are though that the ImageView expects each image to have
the same dimensions.
Simple modification of ImageView example to demonstrate (specify your
image files on command line).
#!/usr/bin/env python3
import numpy as np
from pyqtgraph.Qt import QtCore, QtGui
import pyqtgraph as pg
from argparse import ArgumentParser
from scipy.ndimage import imread
argparser = ArgumentParser(description='Load images and display using
pyqtgraph ImageView widget.')
argparser.add_argument('input_images', help='Input filename(s) of
images', nargs='+')
args = argparser.parse_args()
imagelist = []
print("Loading {}...".format(filename))
image = imread(filename)
imagelist.append(image)
print('Image shape: {}{}'.format(imagelist[len(imagelist)-1].shape,
'' if imagelist[len(imagelist)-1].shape == imagelist[0].shape else ' *'
))
data=np.stack([ i for i in imagelist if i.shape==imagelist[0].shape ], 0)
print('Warning: Shape of all input files must match. Those marked
with * are skipped.')
# Interpret image data as row-major instead of col-major
pg.setConfigOptions(imageAxisOrder='row-major')
app = QtGui.QApplication([])
## Create window with ImageView widget
win = QtGui.QMainWindow()
win.resize(800,800)
imv = pg.ImageView()
win.setCentralWidget(imv)
win.show()
win.setWindowTitle('pyqtgraph example: ImageView')
## Display the data
imv.setImage(data)
## Set a custom color map
colors = [
(0, 0, 0),
(45, 5, 61),
(84, 42, 55),
(150, 87, 60),
(208, 171, 141),
(255, 255, 255)
]
cmap = pg.ColorMap(pos=np.linspace(0.0, 1.0, 6), color=colors)
imv.setColorMap(cmap)
## Start Qt event loop unless running in interactive mode.
import sys
if (sys.flags.interactive != 1) or not hasattr(QtCore,
QtGui.QApplication.instance().exec_()

Post by Dennis Norton
I'm wondering if there is a convenient way to use the ImageView class
to display images either from a list of files in a directory or from a
large binary file consisting of multiple frames.
Could we create a generator which yields the numpy arrays to display?
Thanks.
--

You received this message because you are subscribed to a topic in the
Google Groups "pyqtgraph" group.
To unsubscribe from this topic, visit
https://groups.google.com/d/topic/pyqtgraph/1PuRof6NA5w/unsubscribe.
To unsubscribe from this group and all its topics, send an email to
To view this discussion on the web visit
https://groups.google.com/d/msgid/pyqtgraph/d1681786-8d9c-4a97-a3b5-78f32223e481%40googlegroups.com
<https://groups.google.com/d/msgid/pyqtgraph/d1681786-8d9c-4a97-a3b5-78f32223e481%40googlegroups.com?utm_medium=email&utm_source=footer>
.
For more options, visit https://groups.google.com/d/optout.

--
-Dennis
--

You received this message because you are subscribed to a topic in the
Google Groups "pyqtgraph" group.
To unsubscribe from this topic, visit
https://groups.google.com/d/topic/pyqtgraph/1PuRof6NA5w/unsubscribe.
To unsubscribe from this group and all its topics, send an email to
To view this discussion on the web visit
https://groups.google.com/d/msgid/pyqtgraph/4a8bb184-c59f-4bbc-ae0d-2c8138fc0f9b%40googlegroups.com
<https://groups.google.com/d/msgid/pyqtgraph/4a8bb184-c59f-4bbc-ae0d-2c8138fc0f9b%40googlegroups.com?utm_medium=email&utm_source=footer>
.
For more options, visit https://groups.google.com/d/optout.

--
-Dennis

[Sebastian Höfer]

The most convenient way I used in the past where memory mapped files. Numpy
supports memmapped arrays that (in most cases) behave just like regular
ndarrays.
https://docs.scipy.org/doc/numpy-1.15.1/reference/generated/numpy.memmap.html

We used it to store raw video data with hundreds/thousands of frames and
display them in an ImageView. Of course you're at the whims of your
Operating system and numpy to handle the on-the-fly loading efficiently,
but it's a quick and easy way to handle big amounts of data without
additional libraries and self-made solutions.

Regards,
Sebastian

Post by Dennis Norton
I'm wondering if there is a convenient way to use the ImageView class to
display images either from a list of files in a directory or from a large
binary file consisting of multiple frames.
Could we create a generator which yields the numpy arrays to display?
Thanks.