python - What is dimension order of numpy shape for image data? -

i using nibabel lib load data nii file. read document of lib @ http://nipy.org/nibabel/gettingstarted.html, , found that

this information available without need load of main image data memory. of course there access image data numpy array

this code load data , shapes

import nibabel nib img = nib.load('example.nii') data = img.get_data() data = np.squeeze(data) data = np.copy(data, order="c") print data.shape 

i got result

128, 128, 64 

what order of data shape? widthxheightxdepth? , input must arranged depth, height, width. use input=data.transpose(2,0,1). right? all

update: found numpy read image order height x width x depth reference http://www.python-course.eu/images/axis.jpeg

ok, here's take:

using scipy.ndimage.imread('img.jpg', mode='rgb'), resulting array have order: (h, w, d) i.e. (height, width, depth) because of terminology numpy uses ndarrays (axis=0, axis=1, axis=2) or analogously (y, x, z) if 1 visualize in 3 dimensions.

# read image in [21]: img = scipy.ndimage.imread('suza.jpg', mode='rgb')  # image shape (h, w, d) in [22]: img.shape out[22]: (634, 1366, 3)  # transpose shape (d, h, w) in [23]: tr_img = img.transpose((-1, 0, 1))      in [23]: tr_img.shape out[23]: (3, 634, 1366) 

---

if consider img_shape tuple,

#  index    (0,   1,    2) img_shape = (634, 1366, 3) # or index  (-3,  -2,  -1) 

choose 1 convenient way remember.

---

ps: should noted libraries tensorflow (almost) follows same convention numpy.

tf.image_decode_jpeg() returns:

a tensor of type uint8. 3-d shape [height, width, channels]