diff --git a/skimage/filter/_tv_denoise.py b/skimage/filter/_tv_denoise.py
index 545f9e09..76affb98 100644
--- a/skimage/filter/_tv_denoise.py
+++ b/skimage/filter/_tv_denoise.py
@@ -1,5 +1,5 @@
 import numpy as np
-from skimage.util import dtype
+from skimage import img_as_float
 
 
 def _tv_denoise_3d(im, weight=100, eps=2.e-4, n_iter_max=200):
@@ -171,7 +171,7 @@ def _tv_denoise_2d(im, weight=50, eps=2.e-4, n_iter_max=200):
         i += 1
     return out
     
-def tv_denoise(im, weight=50, eps=2.e-4, keep_type=False, n_iter_max=200):
+def tv_denoise(im, weight=50, eps=2.e-4, n_iter_max=200):
     """
     Perform total-variation denoising on 2-d and 3-d images
 
@@ -192,11 +192,6 @@ def tv_denoise(im, weight=50, eps=2.e-4, keep_type=False, n_iter_max=200):
 
             (E_(n-1) - E_n) < eps * E_0
 
-    keep_type: bool, optional (False)
-        whether the output has the same dtype as the input array.
-        keep_type is False by default, and the dtype of the output
-        is np.float
-
     n_iter_max: int, optional
         maximal number of iterations used for the optimization.
 
@@ -245,7 +240,7 @@ def tv_denoise(im, weight=50, eps=2.e-4, keep_type=False, n_iter_max=200):
     """
     im_type = im.dtype
     if not im_type.kind == 'f':
-        im = im.astype(np.float)
+        im = img_as_float(im)
 
     if im.ndim == 2:
         out = _tv_denoise_2d(im, weight, eps, n_iter_max)
@@ -254,9 +249,4 @@ def tv_denoise(im, weight=50, eps=2.e-4, keep_type=False, n_iter_max=200):
     else:
         raise ValueError('only 2-d and 3-d images may be denoised with this '
                          'function')
-    if keep_type:
-        return out.astype(im_type)
-    else:
-        if not im_type.kind == 'f':
-            out = dtype.convert(out.astype(im_type), np.float)
-        return out      
+    return out      
diff --git a/skimage/filter/tests/test_tv_denoise.py b/skimage/filter/tests/test_tv_denoise.py
index 47f53251..afee6b02 100644
--- a/skimage/filter/tests/test_tv_denoise.py
+++ b/skimage/filter/tests/test_tv_denoise.py
@@ -2,7 +2,7 @@ import numpy as np
 from numpy.testing import run_module_suite
 
 from skimage import filter, data, color
-from skimage import img_as_uint
+from skimage import img_as_uint, img_as_ubyte
 
 
 class TestTvDenoise():
@@ -29,8 +29,8 @@ class TestTvDenoise():
         # test if the total variation has decreased
         assert (np.sqrt((grad_denoised**2).sum())
                 < np.sqrt((grad**2).sum()) / 2)
-        denoised_lena_int = filter.tv_denoise(img_as_uint(lena),
-                                              weight=60.0, keep_type=True)
+        denoised_lena_int = img_as_uint(filter.tv_denoise(img_as_ubyte(lena),
+                                              weight=60.0))
         assert denoised_lena_int.dtype is np.dtype('uint16')
 
     def test_tv_denoise_float_result_range(self):
@@ -55,13 +55,13 @@ class TestTvDenoise():
         mask += 20 * np.random.randn(*mask.shape)
         mask[mask < 0] = 0
         mask[mask > 255] = 255
-        res = filter.tv_denoise(mask.astype(np.uint8),
-                                weight=100, keep_type=True)
+        res = img_as_ubyte(filter.tv_denoise(mask.astype(np.uint8),
+                                weight=100))
         assert res.std() < mask.std()
         assert res.dtype is np.dtype('uint8')
-        res = filter.tv_denoise(mask.astype(np.uint8), weight=100)
+        res = img_as_ubyte(filter.tv_denoise(mask.astype(np.uint8), weight=100))
         assert res.std() < mask.std()
-        assert res.dtype is not np.dtype('uint8')
+
         # test wrong number of dimensions
         a = np.random.random((8, 8, 8, 8))
         try: