notes: special function/nodes skimage

[tokiAi]

Special skimage functions/nodes:

This is a documentation about functions that are special in skimage (order of documentation have no meaning):

---

• Skimage channel_axis argument need to concidered [If None (default argument), the image is assumed to be a grayscale (single channel) image. Otherwise, this parameter indicates which axis of the array corresponds to channels.]:
  like in:

skimage.transform.rescale(image, scale, order=None, mode='reflect', cval=0, clip=True, preserve_range=False, anti_aliasing=None, anti_aliasing_sigma=None, *, channel_axis=None)

---

• Some functions are with Key Words Only Arguments: only some examples, there are more:

skimage.feature.canny(image, sigma=1.0, low_threshold=None, high_threshold=None, mask=None, use_quantiles=False, *, mode='constant', cval=0.0)

solved it for now in the "codeGenerator" like this: added extra strings [mode=, cval=]

...${outputs[1]} = canny(${inputs[1]}['value'], ${inputs[2]}, ${inputs[3]}, ${inputs[4]}, ${inputs[5]}, ${inputs[6]}, mode=${inputs[7]}, cval=${inputs[8]}) ...

---

• skimage.morphology.disk: input: dtype is the data type of the footprint.

solved for now like:

"dtype": {
          "title": "dtype",
          "dataType": "anyDataType",
          "defaultValue": "np.uint8",
          "tooltip": "The data type of the footprint."
        }

---

• skimage.restoration.denoise_wavelet: PyWavelets needed to be installed for this.

---

• skimage.util.img_as_uint: output: converted image as 16-bit unsigned integer format.

solved for now like:

DELETED

---

• Some function can take multiple different argument types like (as
  example - there are more):

skimage.segmentation.watershed connectivity as default 1 (Type: integer) but more flexibility give a ndarray (it is also used in the documentation of the function):

connectivity: ndarray, optional
An array with the same number of dimensions as image whose non-zero elements indicate neighbors for connection. Following the scipy convention, default is a one-connected array of the dimension of the image.

solved with a list/array:

"connectivity": {
          "title": "connectivity",
          "dataType":  ["integer", "numpy.ndarray"],
          "defaultValue": 1,
          "tooltip": "An array with the same number of dimensions as image whose non-zero elements indicate neighbors for connection. Scalar value '1' means 4-connectivity, and '2' would mean 8-connectivity in a 2D spaace."
        }

---

• skimage.measure.label is a function with dynamic outputs (returns) number:

Returns (from skimage doc.): labels, num

num: int, optional
Number of labels, which equals the maximum label index and is only returned if return_num is True.

solved with as: return_num is allways True

---

• skimage.measure.label2rgb have the image as optional and at the second argument:

skimage.color.label2rgb(label, image=None, colors=None, alpha=0.3, bg_label=0, bg_color=(0, 0, 0), image_alpha=1, kind='overlay', *, saturation=0, channel_axis=-1)

---

• Input argument can be inf (np.inf):

e.g skimage.feature.peak_local_max
solved as anyDataType type: e.g.

"num_peaks": {
          "title": "num_peaks",
          "dataType": "anyDataType",
          "defaultValue": "np.inf",
          "tooltip": "Maximum number of peaks."

or

"p_norm": {
          "title": "p_norm",
          "dataType": "anyDataType",
          "defaultValue": "np.inf",
          "tooltip": "Which Minkowski p-norm to use. Should be in the range [1, inf]. inf corresponds to the Chebyshev distance and 2 to the Euclidean distance."
        }

---

• Input argument can be range(...):

e.g skimage.filters.meijering (Ridge_operators)
solved as anyDataType type: e.g.

"sigmas": {
          "title": "sigmas",
          "dataType": "anyDataType",
          "defaultValue": "range(1, 10, 2)",
          "tooltip": "optional iterable of floats indicating scales for the filter. default from doc: range(1, 10, 2) but possible also np.arange(1, 10, 2)"
        }