skimage2.transform.downscale_local_mean#

skimage2.transform.downscale_local_mean(image, factors, cval=0, clip=True)[source]#

Down-sample N-dimensional image by local averaging.

The image is padded with cval if it is not perfectly divisible by the integer factors.

In contrast to interpolation in skimage.transform.resize and skimage.transform.rescale this function calculates the local mean of elements in each block of size factors in the input image.

Parameters:

image(M[, …]) ndarray: Input image.
factorsarray_like: Array containing down-sampling integer factor along each axis.
cvalfloat, optional: Constant padding value if image is not perfectly divisible by the integer factors.
clipbool, optional: Unused, but kept here for API consistency with the other transforms in this module. (The local mean will never fall outside the range of values in the input image, assuming the provided cval also falls within that range.)

Returns:

imagendarray: Down-sampled image with same number of dimensions as input image. For integer inputs, the output dtype will be float64. See numpy.mean() for details.

Examples

>>> a = np.arange(15).reshape(3, 5)
>>> a
array([[ 0,  1,  2,  3,  4],
       [ 5,  6,  7,  8,  9],
       [10, 11, 12, 13, 14]])
>>> downscale_local_mean(a, (2, 3))
array([[3.5, 4. ],
       [5.5, 4.5]])