gaussian_filter of long videos error

[gabriel-ing]

There appears to be an error with openCV gaussian_filter causing an error with my filtering. With one video with >900 frames, the gaussian_filter errored with:

---

AxisError Traceback (most recent call last)
Cell In[6], line 2
1 cc=vid.clip_contrast()
----> 2 cc = cc.gaussian_filter(3)
3 cc = cc.make_scalebar()
4 cc = cc.convert_to_8bit()

File ~/opt/anaconda3/envs/Test/lib/python3.10/site-packages/SimpliPyTEM/MicroVideo_class.py:1222, in MicroVideo.gaussian_filter(self, kernal)
1220 #print(kernal)
1221 swapped = cv.GaussianBlur(swapped, (kernal,kernal), 0)
-> 1222 swapped = np.swapaxes(swapped, 0, 2)
1224 filtered_object = deepcopy(self)
1225 filtered_object.frames = swapped

File <array_function internals>:200, in swapaxes(*args, **kwargs)

File ~/opt/anaconda3/envs/Test/lib/python3.10/site-packages/numpy/core/fromnumeric.py:594, in swapaxes(a, axis1, axis2)
550 @array_function_dispatch(_swapaxes_dispatcher)
551 def swapaxes(a, axis1, axis2):
552 """
553 Interchange two axes of an array.
554
(...)
592
593 """
--> 594 return _wrapfunc(a, 'swapaxes', axis1, axis2)

File ~/opt/anaconda3/envs/Test/lib/python3.10/site-packages/numpy/core/fromnumeric.py:57, in _wrapfunc(obj, method, *args, **kwds)
54 return _wrapit(obj, method, *args, **kwds)
56 try:
---> 57 return bound(*args, **kwds)
58 except TypeError:
59 # A TypeError occurs if the object does have such a method in its
60 # class, but its signature is not identical to that of NumPy's. This
(...)
64 # Call _wrapit from within the except clause to ensure a potential
65 # exception has a traceback chain.
66 return _wrapit(obj, method, *args, **kwds)

AxisError: axis2: axis 2 is out of bounds for array of dimension 2

[gabriel-ing]

The error here is coming from microvideo:

1221 swapped = cv.GaussianBlur(swapped, (kernal,kernal), 0)
-> 1222 swapped = np.swapaxes(swapped, 0, 2)
1224 filtered_object = deepcopy(self)
1225 filtered_object.frames = swapped

The issue is the cv.GaussianBlur is returning a 2 dimensional array rather than a 3D array. leading to an Axis error

[gabriel-ing]

Further investigation suggests its caused by having too many frames of the video, for example:

arr = np.random.rand(1500,1500,500)
g = cv.GaussianBlur(arr, (3,3), 0)

Does not error, however:

arr = np.random.rand(1500,1500,500)
g = cv.GaussianBlur(arr, (3,3), 0)

Raises an error.

Weirdly, performing this here raised an error rather than returning a 2D array, but I assume that reducing the number of frames will still work. Certainly all the videos with fewer frames I have used have worked fine.

[gabriel-ing]

When I first wrote the function, I filtered each frame individually in a big loop, the same way the median filter works, this presumably is the way to bipass this error. The question is, how big a speed loss is this?

If there is a major speed loss I can always use an exception to catch it, something like:
`

    try:
          swapped = cv.GaussianBlur(swapped, (kernal,kernal), 0)
           swapped = np.swapaxes(swapped, 0, 2)
    except:   ***I would like to put the specific error here but the openCV error I'm getting is unnamed
 
    for frame in frames:
       GaussianBlur(frame)

`
etc...