Iterating over a numpy array and operating on each element

Question

I have a numpy array of size 8x8. Here is the numpy array:

QuantTable = np.array([[16, 11 ,10, 16, 24, 40, 51, 61],
                    [12, 12, 14, 19, 26, 58, 60, 55],
                    [14, 13, 16, 24, 40, 57, 69, 56],
                    [14, 17, 22, 29, 51, 87, 80, 62],
                    [18, 22, 37, 29, 51, 87, 80, 62],
                    [24, 35, 55, 64, 81, 109, 103, 77],
                    [49, 64, 78, 87, 103, 121, 120, 101],
                    [72, 92, 95, 98, 112, 100, 103, 99]])

I would like to perform the operations on the elements in the array. I have created a function that accepts a scaling factor value and a Numpy Array. Here it is:

def quantizationTable(Qval, QuantTable):
    if Qval < 50:
        scalingFactor = 5000/Qval
        for x in range(QuantTable):
            for y in range(QuantTable):
                QuantTable[x][y] = ((scalingFactor * QuantTable[x][y] + 50/100)
                if QuantTable[x][y] == 0:
                    QuantTable[x][y] = 1
    return QuantTable
    else:
        scalingFactor = 200 - 2(Qval)
        for x in range(QuantTable):
            for y in range(QuantTable):
                QuantTable[x][y] = ((scalingFactor * QuantTable[x][y] + 50/100)
                if QuantTable[x][y] == 0:
                    QuantTable[x][y] = 1

return QuantTable

I am having trouble iterating over the numpy array and performing my operations. I am trying to apply the formula ((Scaling factor value * element of numpy array + 50)/100) to every element of the numpy array and return the modified array. Can someone please help?

I am having trouble iterating over the numpy array and performing my operations. - Elaborate? — Divakar
– Divakar, Commented Aug 25, 2017 at 16:43
Because I need to perform the operation on each and every element. Please correct me if I am wrong in my logic itself. Thanks. — TheTank
– TheTank, Commented Aug 25, 2017 at 16:46
You don't need to iterate yourself, numpy can do this much faster than any loop in Python could do. — Feodoran
– Feodoran, Commented Aug 25, 2017 at 16:50
@SanketWagh no - indeed, using loops is the inadvisable way of achieving what you want. The whole point of numpy arrays is that they allow vectorized operations that are fast and efficient. — juanpa.arrivillaga
– juanpa.arrivillaga, Commented Aug 25, 2017 at 16:54

cs95 · Accepted Answer · 2017-08-25 16:54:04Z

3

Just remove the loops, and the indexing. Numpy automatically broadcasts those operations. Also, a lot of your code can be taken out of the if...else statements.

def quantizationTable(Qval, QuantTable):
    QuantTable = np.asarray(QuantTable, dtype=np.float32)
    if int(Qval) < 50:
        scalingFactor = 5000 / Qval
    else:
        scalingFactor = 200 - 2 * Qval # confirm that this is what you want? 

    QuantTable *= scalingFactor + 0.5
    QuantTable[QuantTable == 0] = 1

    return QuantTable

edited Aug 25, 2017 at 16:54

answered Aug 25, 2017 at 16:45

cs95

406k106 gold badges744 silver badges797 bronze badges

Sign up to request clarification or add additional context in comments.

4 Comments

MSeifert Over a year ago

You need to be careful with in-place operations. If QuantTable is an integer array that would lead to truncation (because scaling is a float).

cs95 Over a year ago

@MSeifert Oh, good point. I should probably add a np.asarray(..., dtype) cast.

TheTank Over a year ago

scalingFactor = 200 - 2 * Qval -> This is what I want.

TheTank Over a year ago

@COLDSPEED I used this answer and modified it a little. It works like a charm. Thanks.

Collectives™ on Stack Overflow

Iterating over a numpy array and operating on each element

1 Answer 1

4 Comments

Your Answer

Linked

Hot Network Questions

Collectives™ on Stack Overflow

1 Answer 1

4 Comments

Your Answer

Sign up or log in

Post as a guest

Linked

Related