python3-获取600 x 600 numpy数组中nxn数组的平均值

Question

I am writing a method which takes 4 arguments:

1. im: a 600 x 600 numpy array filled with values ranging from 0 - 1000
2. pos_1: x index in im
3. pos_2: y index in im
4. grid_size: size of a sub-grid created given user input, used for averaging

I want to return the average in the nxn grid around the location of [pos_1, pos_2]. Grid_size will be 5, 7 or 9.

Eg. If grid_size was 5, pos_1 was 20 and pos_2 was 150, I would sum the values in a 5 x 5 grid centred around im[20,150] divided by 25 (5 x 5). The current implementation is:

def calc_density(im, pos_1, pos_2, grid_size):
    grid_sum = 0
    if(pos_1 < 600 - (grid_size - 1)/2):
        if(pos_1 > (grid_size - 1)/2):
            if(pos_2 < 600 - (grid_size - 1)/2):
                if(pos_1 > (grid_size - 1)/2):
                    for i in range(grid_size):
                        for j in range(grid_size):
                            grid_sum = grid_sum + im[(pos_1 - (grid_size - 1)/2) + i, (pos_2 - (grid_size - 1)/2) + j]

This method should work for any case where [pos_1, pos_2] does not fall within (grid_size - 1)/2 from the edge of im. If it does then the averaging nxn grid will fall outside of im. If this happens then I am hoping to drop those positions which fall out of im and find the average of those which do fall in the nxn grid.

I can only imagine doing this with a lot of 'if' cases. Is there a better way to do this?

Answer 1

Use numpy facilities for a more readable and more efficient way:

def density(arr,x0,y0,size):
    xb=x0-size//2
    yb=y0-size//2
    return arr[xb:,yb:][:size,:size].mean()

This correctly manages borders in the sense you want :

In [46]: density(arr,603,603,9)
Out[46]: 743.0

In [47]: arr
Out[47]: 
array([[ 73, 197, 311, ..., 952, 477, 138],
       [751,  93, 291, ..., 983, 167, 599],
       [ 54, 666, 380, ..., 456, 466, 754],
       ..., 
       [186, 737, 829, ..., 929,  28, 923],
       [136, 408, 193, ..., 844, 649, 927],
       [477, 411, 458, ...,  64, 173, 743]])