crop center portion of a numpy image

Question

Let’s say I have a numpy image of some width x and height y.
I have to crop the center portion of the image to width cropx and height cropy. Let’s assume that cropx and cropy are positive non zero integers and less than the respective image size. What’s the best way to apply the slicing for the output image?

Asked By: Gert Gottschalk

||

Source

Answer 1

Something along these lines –

def crop_center(img,cropx,cropy):
    y,x = img.shape
    startx = x//2-(cropx//2)
    starty = y//2-(cropy//2)    
    return img[starty:starty+cropy,startx:startx+cropx]

Sample run –

In [45]: img
Out[45]: 
array([[88, 93, 42, 25, 36, 14, 59, 46, 77, 13, 52, 58],
       [43, 47, 40, 48, 23, 74, 12, 33, 58, 93, 87, 87],
       [54, 75, 79, 21, 15, 44, 51, 68, 28, 94, 78, 48],
       [57, 46, 14, 98, 43, 76, 86, 56, 86, 88, 96, 49],
       [52, 83, 13, 18, 40, 33, 11, 87, 38, 74, 23, 88],
       [81, 28, 86, 89, 16, 28, 66, 67, 80, 23, 95, 98],
       [46, 30, 18, 31, 73, 15, 90, 77, 71, 57, 61, 78],
       [33, 58, 20, 11, 80, 25, 96, 80, 27, 40, 66, 92],
       [13, 59, 77, 53, 91, 16, 47, 79, 33, 78, 25, 66],
       [22, 80, 40, 24, 17, 85, 20, 70, 81, 68, 50, 80]])

In [46]: crop_center(img,4,6)
Out[46]: 
array([[15, 44, 51, 68],
       [43, 76, 86, 56],
       [40, 33, 11, 87],
       [16, 28, 66, 67],
       [73, 15, 90, 77],
       [80, 25, 96, 80]])

Answered By: Divakar

Answer 2

Thanks, Divakar.

Your answer got me going the right direction. I came up with this using negative slice offsets to count ‘from the end’:

def cropimread(crop, xcrop, ycrop, fn):
    "Function to crop center of an image file"
    img_pre= msc.imread(fn)
    if crop:
        ysize, xsize, chan = img_pre.shape
        xoff = (xsize - xcrop) // 2
        yoff = (ysize - ycrop) // 2
        img= img_pre[yoff:-yoff,xoff:-xoff]
    else:
        img= img_pre
    return img

Answered By: Gert Gottschalk

Answer 3

A more general solution based on @Divakar ‘s answer:

def cropND(img, bounding):
    start = tuple(map(lambda a, da: a//2-da//2, img.shape, bounding))
    end = tuple(map(operator.add, start, bounding))
    slices = tuple(map(slice, start, end))
    return img[slices]

and if we have an array a

>>> a = np.arange(100).reshape((10,10))

array([[ 0,  1,  2,  3,  4,  5,  6,  7,  8,  9],
       [10, 11, 12, 13, 14, 15, 16, 17, 18, 19],
       [20, 21, 22, 23, 24, 25, 26, 27, 28, 29],
       [30, 31, 32, 33, 34, 35, 36, 37, 38, 39],
       [40, 41, 42, 43, 44, 45, 46, 47, 48, 49],
       [50, 51, 52, 53, 54, 55, 56, 57, 58, 59],
       [60, 61, 62, 63, 64, 65, 66, 67, 68, 69],
       [70, 71, 72, 73, 74, 75, 76, 77, 78, 79],
       [80, 81, 82, 83, 84, 85, 86, 87, 88, 89],
       [90, 91, 92, 93, 94, 95, 96, 97, 98, 99]])

We can clip it with cropND(a, (5,5)), you will get:

>>> cropND(a, (5,5))

array([[33, 34, 35, 36, 37],
       [43, 44, 45, 46, 47],
       [53, 54, 55, 56, 57],
       [63, 64, 65, 66, 67],
       [73, 74, 75, 76, 77]])

It not only works with 2D image but also 3D image.

Have a nice day.

Answered By: Losses Don

Answer 4

A simple modification from @Divakar ‘s answer that preserves the image channel:

    def crop_center(self, img, cropx, cropy):
       _, y, x = img.shape
       startx = x // 2 - (cropx // 2)
       starty = y // 2 - (cropy // 2)
       return img[:, starty:starty + cropy, startx:startx + cropx]

Answered By: kdebugging

Answer 5

Another simple modification from @Divakar’s answer to preserve color channels:

def crop_center(img,cropx,cropy):
    y,x,_ = img.shape
    startx = x//2-(cropx//2)
    starty = y//2-(cropy//2)
    return img[starty:starty+cropy,startx:startx+cropx,:]

Answered By: leenremm

Answer 6

I had a problem in which I needed to crop the center from both 2D and 3D arrays. That meant unpacking a varying number of items from img.shape.

Dropping this here in case someone runs into the same problem. This modification of Divkar’s solution allows cropping 2D OR 3D arrays

def crop_center(img, cropx, cropy):
    y, x, *_ = img.shape
    startx = x // 2 - (cropx // 2)
    starty = y // 2 - (cropy // 2)    
    return img[starty:starty + cropy, startx:startx + cropx, ...]

Answered By: JmeCS

crop center portion of a numpy image

Question:

Answers: