diff --git a/bento.info b/bento.info
index cba403d9..7d21bfa8 100644
--- a/bento.info
+++ b/bento.info
@@ -161,6 +161,9 @@ Library:
     Extension: skimage.external.tifffile._tifffile
         Sources:
             skimage/external/tifffile/_tifffile.c
+    Extension: skimage.transform._seam_carving
+        Sources:
+            skimage/transform/_seam_carving.pyx
 
 Executable: skivi
     Module: skimage.scripts.skivi
diff --git a/doc/examples/plot_seam_carving.py b/doc/examples/plot_seam_carving.py
new file mode 100644
index 00000000..f1dcb948
--- /dev/null
+++ b/doc/examples/plot_seam_carving.py
@@ -0,0 +1,95 @@
+"""
+============
+Seam Carving
+============
+
+This example demonstrates how images can be resized using seam carving [1]_.
+Resizing to a new aspect ratio distorts image contents. Seam carving attempts
+to resize *without* distortion, by removing regions of an image which are less
+important. In this example we are using the Sobel filter to signify the
+importance of each pixel.
+
+.. [1] Shai Avidan and Ariel Shamir
+       "Seam Carving for Content-Aware Image Resizing"
+       http://www.cs.jhu.edu/~misha/ReadingSeminar/Papers/Avidan07.pdf
+
+"""
+from skimage import data, draw
+from skimage import transform, util
+import numpy as np
+from skimage import filters, color
+from matplotlib import pyplot as plt
+
+
+hl_color = np.array([0, 1, 0])
+
+img = data.rocket()
+img = util.img_as_float(img)
+eimg = filters.sobel(color.rgb2gray(img))
+
+plt.title('Original Image')
+plt.imshow(img)
+
+"""
+.. image:: PLOT2RST.current_figure
+"""
+
+resized = transform.resize(img, (img.shape[0], img.shape[1] - 200))
+plt.figure()
+plt.title('Resized Image')
+plt.imshow(resized)
+
+
+"""
+.. image:: PLOT2RST.current_figure
+"""
+
+out = transform.seam_carve(img, eimg, 'vertical', 200)
+plt.figure()
+plt.title('Resized using Seam-Carving')
+plt.imshow(out)
+
+"""
+.. image:: PLOT2RST.current_figure
+
+As you can see, resizing as distorted the rocket and the objects around,
+whereas seam carving has reszied by removing the empty spaces in between.
+
+Object Removal
+--------------
+
+Seam Carving can also be used to remove atrifacts from images. To do that, we
+have to ensure that pixels to be removes get less importance. In the following
+code I approximately mark the rocket with a mask, and then decrease the
+importance of those pixels
+
+"""
+
+masked_img = img.copy()
+
+poly = [(404, 281), (404, 360), (359, 364), (338, 337), (145, 337), (120, 322),
+        (145, 304), (340, 306), (362, 284)]
+pr = np.array([p[0] for p in poly])
+pc = np.array([p[1] for p in poly])
+rr, cc = draw.polygon(pr, pc)
+
+masked_img[rr, cc, :] = masked_img[rr, cc, :]*0.5 + hl_color*.5
+plt.figure()
+plt.title('Object Marked')
+
+plt.imshow(masked_img)
+"""
+.. image:: PLOT2RST.current_figure
+"""
+
+eimg[rr, cc] -= 1000
+
+plt.figure()
+plt.title('Object Removed')
+out = transform.seam_carve(img, eimg, 'vertical', 90)
+resized = transform.resize(img, out.shape)
+plt.imshow(out)
+plt.show()
+"""
+.. image:: PLOT2RST.current_figure
+"""
diff --git a/skimage/data/__init__.py b/skimage/data/__init__.py
index 1711185c..a5f5068d 100644
--- a/skimage/data/__init__.py
+++ b/skimage/data/__init__.py
@@ -26,6 +26,7 @@ __all__ = ['load',
            'chelsea',
            'coffee',
            'hubble_deep_field',
+           'rocket',
            'astronaut']
 
 
@@ -241,3 +242,22 @@ def hubble_deep_field():
 
     """
     return load("hubble_deep_field.jpg")
+
+
+def rocket():
+    """Launch photo of DSCOVR on Falcon 9 by SpaceX.
+
+    This is the launch photo of Falcon 9 carrying DSCOVR lifted off from
+    SpaceX's Launch Complex 40 at Cape Canaveral Air Force Station, FL.
+
+    Notes
+    -----
+    This image was downloaded from
+    `SpaceX Photos
+    <https://www.flickr.com/photos/spacexphotos/16511594820/in/photostream/>`__.
+
+    The image was captured by SpaceX and `released in the public domain
+    <http://arstechnica.com/tech-policy/2015/03/elon-musk-puts-spacex-photos-into-the-public-domain/>`_.
+
+    """
+    return load("rocket.jpg")
diff --git a/skimage/data/rocket.jpg b/skimage/data/rocket.jpg
new file mode 100644
index 00000000..32bdd339
Binary files /dev/null and b/skimage/data/rocket.jpg differ
diff --git a/skimage/transform/__init__.py b/skimage/transform/__init__.py
index 3049ebb0..9b40b3a9 100644
--- a/skimage/transform/__init__.py
+++ b/skimage/transform/__init__.py
@@ -12,6 +12,7 @@ from ._geometric import (warp, warp_coords, estimate_transform,
 from ._warps import swirl, resize, rotate, rescale, downscale_local_mean
 from .pyramids import (pyramid_reduce, pyramid_expand,
                        pyramid_gaussian, pyramid_laplacian)
+from .seam_carving import seam_carve
 
 
 __all__ = ['hough_circle',
@@ -43,4 +44,5 @@ __all__ = ['hough_circle',
            'pyramid_reduce',
            'pyramid_expand',
            'pyramid_gaussian',
-           'pyramid_laplacian']
+           'pyramid_laplacian',
+           'seam_carve']
diff --git a/skimage/transform/_seam_carving.pyx b/skimage/transform/_seam_carving.pyx
new file mode 100644
index 00000000..3499514c
--- /dev/null
+++ b/skimage/transform/_seam_carving.pyx
@@ -0,0 +1,226 @@
+# cython: cdivision=True
+# cython: boundscheck=False
+# cython: nonecheck=False
+# cython: wraparound=False
+import numpy as np
+cimport numpy as cnp
+
+
+cdef cnp.double_t DBL_MAX = np.finfo(np.double).max
+
+cdef void _preprocess_image(cnp.double_t[:, :, ::1] energy_img,
+                            cnp.double_t[:, ::1] cumulative_img,
+                            cnp.int8_t[:, ::1] track_img,
+                            Py_ssize_t cols) nogil:
+    """ For each row, compute the lowest seam value for all its columns.
+
+    This function updates `cumulative_img` such that `cumulative_img[r, c]`
+    is the total energy of the lowest energy seam ending at `(r, c)`.
+
+    Parameters
+    ----------
+    energy_img : (M, N, 1) ndarray
+        Cost array representing the expense to remove each pixel. Seam carving
+        tries to avoid pixels with high costs.
+    cumulative_img : (M, N) ndarray
+        The array to be updated inplace with the total cost of lowest energy
+        seams.
+    track_img : (M, N) ndarray
+        For each pixel, `track_img` stores the relative column offset in
+        the previous row which has the lowest value in `cumulative_img`. This
+        helps in in re-tracing the minimum cost seam.
+    cols : int
+        Number of columns to process.
+    """
+
+    cdef Py_ssize_t r, c, offset, c_idx
+    cdef Py_ssize_t rows = energy_img.shape[0]
+    cdef cnp.double_t min_cost = DBL_MAX
+    cdef Py_ssize_t colsm1 = cols - 1
+    cdef Py_ssize_t rm1
+
+    for c in range(cols):
+        cumulative_img[0, c] = energy_img[0, c, 0]
+
+    for r in range(1, rows):
+        rm1 = r - 1
+        for c in range(cols):
+            min_cost = DBL_MAX
+            for offset in range(-1, 2):
+
+                c_idx = c + offset
+                if (c_idx > colsm1) or (c_idx < 0):
+                    continue
+
+                if cumulative_img[rm1, c_idx] < min_cost:
+                    min_cost = cumulative_img[rm1, c_idx]
+                    track_img[r, c] = offset
+
+            cumulative_img[r, c] = min_cost + energy_img[r, c, 0]
+
+cdef bint _mark_seam(cnp.int8_t[:, ::1] track_img,
+                     Py_ssize_t start_index,
+                     cnp.uint8_t[:, ::1] seam_map,
+                     Py_ssize_t[::1] seam_buffer) nogil:
+
+    """ Re-trace the optimal seam from a given column in the last row.
+
+    This function tries to re-track an optimal seam from `start_index` and
+    tries to mark it in `seam_map`. If this seam intersects with any existing
+    seam in `seam_map` the function returns `0` without marking anything. Else
+    it marks the seam in `seam_map` and returns `1`.
+
+    track_img : (M, N) ndarray
+       The array of relative column indices as updated by `_preprocess_image`.
+    start_index : int
+       The column number of the bottom most row from where to start re-tracing
+       the seam.
+    seam_map : (M, N) ndarray
+        The array used to mark seams. If a pixel is marked as as seam it is set
+        to `1`, else `0`.
+    seam_buffer : (M,) ndarray
+        Buffer used to store the column indices of the seam currently being
+        checked. This is preallocated to save time.
+
+    Returns
+    -------
+    success : int
+        `1` if seam was marked, `0` is seam intersects and was not marked.
+    """
+    cdef Py_ssize_t rows = track_img.shape[0]
+    cdef Py_ssize_t[::1] current_seam_indices = seam_buffer
+    cdef Py_ssize_t row, col
+    cdef cnp.int8_t offset
+    cdef Py_ssize_t seams
+
+    current_seam_indices[rows - 1] = start_index
+    for row in range(rows - 2, -1, -1):
+        col = current_seam_indices[row + 1]
+        offset = track_img[row, col]
+        col = col + offset
+        current_seam_indices[row] = col
+
+        if seam_map[row, col]:
+            return 0
+
+    for row in range(rows):
+        col = current_seam_indices[row]
+        seam_map[row, col] = 1
+
+    return 1
+
+cdef void _remove_seam(cnp.double_t[:, :, ::1] img,
+                       cnp.uint8_t[:, ::1] seam_map, Py_ssize_t cols) nogil:
+    """ Remove marked seams from an image.
+
+    Parameters
+    ----------
+    img : (M, N, P) ndarray
+        Input image whose vertical seams are to be removed.
+    seam_map : (M, N) ndarray
+        Array with seams to be removed marked by non-zero entries.
+    cols : int
+        The number of columns to process.
+    """
+    cdef Py_ssize_t rows = img.shape[0]
+    cdef Py_ssize_t channels = img.shape[2]
+    cdef Py_ssize_t r, c, ch, shift
+    cdef Py_ssize_t c_shift
+
+    for r in range(rows):
+        shift = 0
+        for c in range(cols):
+            shift += seam_map[r, c]
+            c_shift = c + shift
+            for ch in range(channels):
+                img[r, c, ch] = img[r, c_shift, ch]
+
+
+def _seam_carve_v(img, energy_map, iters, border):
+    """ Carve vertical seams off an image.
+
+    Carves out vertical seams from an image while using the given energy map to
+    decide the importance of each pixel.[1]_
+
+    Parameters
+    ----------
+    img : (M, N) or (M, N, 3) ndarray
+        Input image whose vertical seams are to be removed.
+    energy_map : (M, N) ndarray
+        Cost array denoting importance of each pixel. The algorithm will try to
+        retain high valued pixels.
+    iters : int
+        Number of vertical seams to be removed.
+    border : int, optional
+        The number of pixels in the right, left and bottom end of the image
+        to be excluded from being considered for a seam. This is important as
+        certain filters just ignore image boundaries and set them to `0`.
+        By default border is set to `1`.
+
+    Returns
+    -------
+    image : (M, N - iters, 3) ndarray of float
+        The cropped image with the vertical seams removed.
+
+    References
+    ----------
+    .. [1] Shai Avidan and Ariel Shamir
+           "Seam Carving for Content-Aware Image Resizing"
+           http://www.cs.jhu.edu/~misha/ReadingSeminar/Papers/Avidan07.pdf
+    """
+    # This reference has been kept to be used for the `np.argsort` call
+    last_row_obj = np.zeros(img.shape[1], dtype=np.float)
+
+    cdef cnp.double_t[::1] last_row = last_row_obj
+    cdef Py_ssize_t[::1] sorted_indices
+    cdef cnp.uint8_t[:, ::1] seam_map = np.zeros(img.shape[0:2],
+                                                 dtype=np.uint8)
+    cdef Py_ssize_t cols = img.shape[1]
+    cdef Py_ssize_t rows = img.shape[0]
+    cdef Py_ssize_t seams_left = iters
+    cdef Py_ssize_t seams_removed
+    cdef Py_ssize_t seam_idx
+    cdef Py_ssize_t[::1] seam_buffer = np.zeros(rows, dtype=np.int)
+
+    cdef cnp.double_t[:, :, ::1] image = img
+    cdef cnp.int8_t[:, ::1] track_img = np.zeros(img.shape[0:2], dtype=np.int8)
+    cdef cnp.double_t[:, ::1] cumulative_img = np.zeros(img.shape[0:2],
+                                                        dtype=np.float)
+    cdef cnp.double_t[:, :, ::1] energy_img
+
+    energy_map[:, 0:border] = DBL_MAX
+    energy_map[:, cols-border:cols] = DBL_MAX
+
+    # Filters often let the boundary be `0`. If all the entries in the last
+    # row of `energy_img` are equal, the minimum value in the penultimate row
+    # of `cumulative_img` will result in 3 minimum values in its last row.
+    # Hence, two successive removals will always intersect as the 3 least seams
+    # will share the same pixels except they will differ in the last row.
+    energy_map[rows-border:rows, :] = energy_map[rows-2*border:rows-border, :]
+
+    energy_map = np.ascontiguousarray(energy_map[:, :, np.newaxis])
+    energy_img = energy_map
+
+    _preprocess_image(energy_img, cumulative_img, track_img, cols)
+    last_row[...] = cumulative_img[rows - 1, :]
+    sorted_indices = np.argsort(last_row_obj)
+    seam_idx = 0
+
+    while seams_left > 0:
+        if _mark_seam(track_img, sorted_indices[seam_idx], seam_map,
+                      seam_buffer):
+            seams_left -= 1
+            cols -= 1
+            seam_idx += 1
+        else:
+            seam_idx = 0
+            _remove_seam(image, seam_map, cols)
+            _remove_seam(energy_img, seam_map, cols)
+            seam_map[...] = 0
+            _preprocess_image(energy_img, cumulative_img, track_img, cols)
+            last_row[:cols] = cumulative_img[rows - 1, :cols]
+            sorted_indices = np.argsort(last_row_obj)
+
+    _remove_seam(image, seam_map, cols)
+
+    return img[:, 0:cols]
diff --git a/skimage/transform/seam_carving.py b/skimage/transform/seam_carving.py
new file mode 100644
index 00000000..95e975ce
--- /dev/null
+++ b/skimage/transform/seam_carving.py
@@ -0,0 +1,66 @@
+from ._seam_carving import _seam_carve_v
+from .. import util
+from .._shared import utils
+import numpy as np
+
+
+def seam_carve(img, energy_map, mode, num, border=1, force_copy=True):
+    """ Carve vertical or horizontal seams off an image.
+
+    Carves out vertical/horizontal seams from an image while using the given
+    energy map to decide the importance of each pixel.
+
+    Parameters
+    ----------
+    image : (M, N) or (M, N, 3) ndarray
+        Input image whose seams are to be removed.
+    energy_map : (M, N) ndarray
+        The array to decide the importance of each pixel. The higher
+        the value corresponding to a pixel, the more the algorithm will try
+        to keep it in the image.
+    mode : str {'horizontal', 'vertical'}
+        Indicates whether seams are to be removed vertically or horizontally.
+        Removing seams horizontally will decrease the height whereas removing
+        vertically will decrease the width.
+    num : int
+        Number of seams are to be removed.
+    border : int, optional
+        The number of pixels in the right, left and bottom end of the image
+        to be excluded from being considered for a seam. This is important as
+        certain filters just ignore image boundaries and set them to `0`.
+        By default border is set to `1`.
+    force_copy : bool, optional
+        If set, the `image` and `energy_map` are copied before being used by
+        the method which modifies it in place. Set this to `False` if the
+        original image and the energy map are no longer needed after
+        this opetration.
+
+    Returns
+    -------
+    out : ndarray
+        The cropped image with the seams removed.
+
+    References
+    ----------
+    .. [1] Shai Avidan and Ariel Shamir
+           "Seam Carving for Content-Aware Image Resizing"
+           http://www.cs.jhu.edu/~misha/ReadingSeminar/Papers/Avidan07.pdf
+    """
+
+    utils.assert_nD(img, (2, 3))
+    image = util.img_as_float(img, force_copy)
+    energy_map = util.img_as_float(energy_map, force_copy)
+
+    if image.ndim == 2:
+        image = image[..., np.newaxis]
+
+    if mode == 'horizontal':
+        image = np.transpose(image, (1, 0, 2))
+
+    image = np.ascontiguousarray(image)
+    out = _seam_carve_v(image, energy_map, num, border)
+
+    if mode == 'horizontal':
+        out = np.transpose(out, (1, 0, 2))
+
+    return np.squeeze(out)
diff --git a/skimage/transform/setup.py b/skimage/transform/setup.py
index 22f31696..ff2e9bc6 100644
--- a/skimage/transform/setup.py
+++ b/skimage/transform/setup.py
@@ -16,6 +16,7 @@ def configuration(parent_package='', top_path=None):
     cython(['_hough_transform.pyx'], working_path=base_path)
     cython(['_warps_cy.pyx'], working_path=base_path)
     cython(['_radon_transform.pyx'], working_path=base_path)
+    cython(['_seam_carving.pyx'], working_path=base_path)
 
     config.add_extension('_hough_transform', sources=['_hough_transform.c'],
                          include_dirs=[get_numpy_include_dirs()])
@@ -27,6 +28,8 @@ def configuration(parent_package='', top_path=None):
                          sources=['_radon_transform.c'],
                          include_dirs=[get_numpy_include_dirs()])
 
+    config.add_extension('_seam_carving', sources=['_seam_carving.c'],
+                         include_dirs=[get_numpy_include_dirs()])
     return config
 
 if __name__ == '__main__':
diff --git a/skimage/transform/tests/test_seam_carving.py b/skimage/transform/tests/test_seam_carving.py
new file mode 100644
index 00000000..b4eb8977
--- /dev/null
+++ b/skimage/transform/tests/test_seam_carving.py
@@ -0,0 +1,23 @@
+from skimage import transform
+import numpy as np
+from numpy import testing
+
+
+def test_seam_carving():
+    img = np.array([[0, 0, 1, 0, 0],
+                    [0, 0, 1, 0, 0],
+                    [0, 0, 1, 0, 0],
+                    [0, 1, 0, 0, 0],
+                    [1, 0, 0, 0, 0]], dtype=np.float)
+    energy = 1 - img
+
+    out = transform.seam_carve(img, energy, 'vertical', 1, border=0)
+    testing.assert_allclose(out, 0)
+
+    img = img.T
+    out = transform.seam_carve(img, energy, 'horizontal', 1, border=0)
+    testing.assert_allclose(out, 0)
+
+
+if __name__ == '__main__':
+    np.testing.run_module_suite()