xoa.filter.generate_kernel

xoa.filter.generate_kernel(kernel, data, window_func='ones', isotropic=False, fill_value=0.0, window_kwargs=None, **kwargs)[source]

Generate a kernel that is compatible with a given data array

Parameters:

  • kernel (xarray.DataArray, np.ndarray, int, list, dictorthokernels) –

    Ready to use kernel or specs to generate it.

    • If an int, the kernel built with ones with a size of kernel along all dimensions.

    • If a tuple, the kernel is built with ones and a shape equal to kernel.

    • If a numpy array, it is used as is.

    The final data array is transposed and/or expanded with xoa.coords.transpose() to fit into the input data array.

  • data (xarray.DataArray) – Data array that the kernel must be compatible with. If the kernel has more than one dimension, it is expanded with a size of 1 for missing dimensions.

  • isotropic (bool, tuple) – Tuple of the dimensions over which must be computed isotropically.

  • fill_value (float) – Value used by generate_isotropic_kernel() to fill the isotropic kernel in its corners.

  • window_func (str) – Function to generate the kernel from its size by calling get_window_func()

  • window_kwargs (dict) – Optional arguments passed to get_window_func()

  • kwargs (dict) – Extra parameters are merged with window_kwargs

Returns:

xarray.DataArray – Kernel array with suitable dimensions and shape

See also

generate_isotropic_kernel, generate_orthogonal_kernel, get_window_func, scipy.signal.windows, numpy.bartlett, numpy.blackman, numpy.hamming, numpy.hanning, numpy.kaiser, xoa.coords.transpose