I have a large numpy array that I need to manipulate so that each element is changed to either a 1 or 0 if a condition is met (will be used as a pixel mask later). There are about 8 million elements in the array and my current method takes too long for the reduction pipeline:
for (y,x), value in numpy.ndenumerate(mask_data): if mask_data[y,x]<3: #Good Pixel mask_data[y,x]=1 elif mask_data[y,x]>3: #Bad Pixel mask_data[y,x]=0 Is there a numpy function that would speed this up?
>>> import numpy as np >>> a = np.random.randint(0, 5, size=(5, 4)) >>> a array([[4, 2, 1, 1], [3, 0, 1, 2], [2, 0, 1, 1], [4, 0, 2, 3], [0, 0, 0, 2]]) >>> b = a < 3 >>> b array([[False, True, True, True], [False, True, True, True], [ True, True, True, True], [False, True, True, False], [ True, True, True, True]], dtype=bool) >>> >>> c = b.astype(int) >>> c array([[0, 1, 1, 1], [0, 1, 1, 1], [1, 1, 1, 1], [0, 1, 1, 0], [1, 1, 1, 1]]) You can shorten this with:
>>> c = (a < 3).astype(int) 
>>> a = np.random.randint(0, 5, size=(5, 4)) >>> a array([[0, 3, 3, 2], [4, 1, 1, 2], [3, 4, 2, 4], [2, 4, 3, 0], [1, 2, 3, 4]]) >>> >>> a[a > 3] = -101 >>> a array([[ 0, 3, 3, 2], [-101, 1, 1, 2], [ 3, -101, 2, -101], [ 2, -101, 3, 0], [ 1, 2, 3, -101]]) >>> See, eg, Indexing with boolean arrays.
a[a > 3] = -101+a[a > 3].
a[a > 3] = -101+a[a > 3] instead of a[a > 3] += -101 you will most likely face memory leakage.
The quickest (and most flexible) way is to use np.where, which chooses between two arrays according to a mask(array of true and false values):
import numpy as np a = np.random.randint(0, 5, size=(5, 4)) b = np.where(a<3,0,1) print('a:',a) print() print('b:',b) which will produce:
a: [[1 4 0 1] [1 3 2 4] [1 0 2 1] [3 1 0 0] [1 4 0 1]] b: [[0 1 0 0] [0 1 0 1] [0 0 0 0] [1 0 0 0] [0 1 0 0]] 
a[a<3] = 0You can create your mask array in one step like this
mask_data = input_mask_data < 3 This creates a boolean array which can then be used as a pixel mask. Note that we haven't changed the input array (as in your code) but have created a new array to hold the mask data - I would recommend doing it this way.
>>> input_mask_data = np.random.randint(0, 5, (3, 4)) >>> input_mask_data array([[1, 3, 4, 0], [4, 1, 2, 2], [1, 2, 3, 0]]) >>> mask_data = input_mask_data < 3 >>> mask_data array([[ True, False, False, True], [False, True, True, True], [ True, True, False, True]], dtype=bool) >>> 
.astype(int) or *1, but an array of True and False is just as good as it is.I was a noob with Numpy, and the answers above where not straight to the point to modify in place my array, so I'm posting what I came up with:
import numpy as np arr = np.array([[[10,20,30,255],[40,50,60,255]], [[70,80,90,255],[100,110,120,255]], [[170,180,190,255],[230,240,250,255]]]) # Change 1: # Set every value to 0 if first element is smaller than 80 arr[arr[:,:,0] < 80] = 0 print('Change 1:',arr,'\n') # Change 2: # Set every value to 1 if bigger than 180 and smaller than 240 # OR if equal to 170 arr[(arr > 180) & (arr < 240) | (arr == 170)] = 1 print('Change 2:',arr) This produces:
Change 1: [[[ 0 0 0 0] [ 0 0 0 0]] [[ 0 0 0 0] [100 110 120 255]] [[170 180 190 255] [230 240 250 255]]] Change 2: [[[ 0 0 0 0] [ 0 0 0 0]] [[ 0 0 0 0] [100 110 120 255]] [[ 1 180 1 255] [ 1 240 250 255]]] This way you can add tons of conditions like 'Change 2' and set values accordingly.
I am not sure I understood your question, but if you write:
mask_data[:3, :3] = 1 mask_data[3:, 3:] = 0 This will make all values of mask data whose x and y indexes are less than 3 to be equal to 1 and all rest to be equal to 0
mask_data[y,x]==3?if mask_data[y,x]>=3: