Elastic transformation of an image in Python

elastic_transform.py

import numpy
from scipy.ndimage.interpolation import map_coordinates
from scipy.ndimage.filters import gaussian_filter

def elastic_transform(image, alpha, sigma, random_state=None):
    """Elastic deformation of images as described in [Simard2003]_.


    .. [Simard2003] Simard, Steinkraus and Platt, "Best Practices for
       Convolutional Neural Networks applied to Visual Document Analysis", in
       Proc. of the International Conference on Document Analysis and
       Recognition, 2003.
    """
    if random_state is None:
        random_state = numpy.random.RandomState(None)

    shape = image.shape
    dx = gaussian_filter((random_state.rand(*shape) * 2 - 1), sigma, mode="constant", cval=0) * alpha
    dy = gaussian_filter((random_state.rand(*shape) * 2 - 1), sigma, mode="constant", cval=0) * alpha

    x, y = numpy.meshgrid(numpy.arange(shape[0]), numpy.arange(shape[1]))
    indices = numpy.reshape(y+dy, (-1, 1)), numpy.reshape(x+dx, (-1, 1))

    return map_coordinates(image, indices, order=1).reshape(shape)

   indices = numpy.reshape(y+dy, (-1, 1)), numpy.reshape(x+dx, (-1, 1))
ValueError: operands could not be broadcast together with shapes (580,420) (420,580) 

It seems that we cannot do y+dy since shape does not match?

I do believe line 24 has it's shape values reversed. Should be:
x, y = numpy.meshgrid(numpy.arange(shape[1]), numpy.arange(shape[0]))

Unless the image is square in which case it doesn't matter.

Above script doesn't work if image is not a square.
Here's a working script: https://gist.github.com/chsasank/4d8f68caf01f041a6453e67fb30f8f5a

what code did you use to load in the "image"?

@chethanjjj I use Image from PIL
You can install it with
pip install Pillow=4.2.1

Thanks for sharing this.

Any possibility you could use a license so I know if and how I can use it?

thanks for sharing.
Is it code working with dicom images?

A c++ implementation can be found here Implementation of elastic distortion algorithm in C++