read dv videos

bigfish/stack/__init__.py

@@ -6,8 +6,10 @@
 """
 
 from .utils import (check_array, check_df, check_recipe, check_parameter,
-                    check_range_value, get_offset_value, get_eps_float32)
+                    check_range_value,
+                    get_offset_value, get_eps_float32)
 from .io import (read_image, read_pickle, read_cell_json, read_rna_json,
+                 read_dv,
                  save_image)
 from .preprocess import (build_simulated_dataset, build_stacks, build_stack,
                          build_stack_no_recipe, rescale,
@@ -43,6 +45,7 @@
           "check_range_value", "get_offset_value", "get_eps_float32"]
 
 _io = ["read_image", "read_pickle", "read_cell_json", "read_rna_json",
+       "read_dv",
        "save_image"]
 
 _preprocess = ["build_simulated_dataset", "build_stacks", "build_stack",

bigfish/stack/io.py

@@ -6,6 +6,7 @@
 """
 
 import pickle
+import mrc
 import warnings
 
 import numpy as np
@@ -21,7 +22,7 @@ def read_image(path):
     """Read an image with the .png, .tif or .tiff extension.
 
     The input image should be in 2-d or 3-d, with unsigned integer 8 or 16
-    bits, integer
+    bits.
 
     Parameters
     ----------
@@ -118,7 +119,7 @@ def read_pickle(path):
 
     Returns
     -------
-    data = pandas.DataFrame or np.ndarray
+    data : pandas.DataFrame or np.ndarray
         Data store in the pickle file (an image or coordinates with labels and
         metadata).
 
@@ -130,6 +131,35 @@ def read_pickle(path):
     return data
 
 
+def read_dv(path):
+    """Read a video file with the .dv extension.
+
+    The input image should be in 2-d or 3-d, with unsigned integer 8 or 16
+    bits.
+
+    Parameters
+    ----------
+    path : str
+        Path of the file to read.
+
+    Returns
+    -------
+    tensor : ndarray, np.uint or np.int
+        A 2-d or 3-d tensor with spatial dimensions.
+    """
+    # TODO allow more input dtype
+    # read video file
+    tensor = mrc.imread(path)
+
+    # check the image is in unsigned integer 16 bits with 2 or 3 dimensions
+    check_array(tensor,
+                dtype=[np.uint8, np.uint16, np.int64],
+                ndim=[2, 3],
+                allow_nan=False)
+
+    return tensor
+
+
 # ### Write ###
 
 def save_image(image, path):