Efficient element-wise multiplication in Python with numpy array

Question

I have a numpy array of type np.int64, to which I am trying to apply a formula.

Let's say the array is a 2D array called total_img which has dimensions 400 X 300 pixels. For each pixel I want to compute the following formula. px = 255*(min - px)/(min - max). I want these rescaled pixels to be stored always in total_img. How can I efficiently implement this using numpy arrays?

Note, min and max are simply the 1th percentile and 99th percentile values and are actually stored as floats. Should I convert them to ints for better accuracy (remember total_img is of type np.int64 - No Overflow will ever occur). Also min will most likely be negative.

I was using:

for row in total_img:
    for px in row:        
        px = 255*(min-px)/(min - max)

you should not shadow builtins such as min/max ... just an aside totally unrelated to your problem — Joran Beasley
– Joran Beasley, Commented Jun 24, 2015 at 23:00

user2357112 · Accepted Answer · 2015-06-24 23:00:58Z

3

total_img = 255*(min - total_img)/(min - max)

You literally just plug in total_img instead of px, and it does the operation elementwise. If you want to store the result into the original total_img instead of replacing it, you can slice-assign:

total_img[:] = 255*(min - total_img)/(min - max)

but note that this doesn't actually save any time or memory.

answered Jun 24, 2015 at 23:00

user2357112

286k32 gold badges490 silver badges570 bronze badges

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

Comments

sparklearner · Accepted Answer · 2015-06-24 23:04:55Z

0

I believe you could directly do this:

total_img = 255*(min-total_img)/(min - max)

answered Jun 24, 2015 at 23:04

sparklearner

4033 silver badges7 bronze badges

Collectives™ on Stack Overflow

Efficient element-wise multiplication in Python with numpy array

2 Answers 2

Comments

Comments

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

2 Answers 2

Comments

Comments

Your Answer

Sign up or log in

Post as a guest

Related