OpenCV Python binds incredibly slow iterations through image data

Question

I recently took some code that tracked an object based on color in OpenCV c++ and rewrote it in the python bindings.

The overall results and method were the same minus syntax obviously. But, when I perform the below code on each frame of a video it takes almost 2-3 seconds to complete where as the c++ variant, also below, is instant in comparison and I can iterate between frames as fast as my finger can press a key.

Any ideas or comments?

    cv.PyrDown(img, dsimg)
    for i in range( 0, dsimg.height ):
        for j in range( 0, dsimg.width):

            if dsimg[i,j][1] > ( _RED_DIFF + dsimg[i,j][2] ) and dsimg[i,j][1] > ( _BLU_DIFF + dsimg[i,j][0] ):
                res[i,j] = 255
            else:
                res[i,j] = 0

    for( int i =0; i < (height); i++ ) 
    {
        for( int j = 0; j < (width); j++ )
        {
            if( ( (data[i * step + j * channels + 1]) > (RED_DIFF + data[i * step + j * channels + 2]) ) &&
                ( (data[i * step + j * channels + 1]) > (BLU_DIFF + data[i * step + j * channels]) ) )
                data_r[i *step_r + j * channels_r] = 255;
            else
                data_r[i * step_r + j * channels_r] = 0;
        }
    }

Thanks

Joe Kington · Accepted Answer · 2010-08-22 20:44:24Z

6

Try using numpy to do your calculation, rather than nested loops. You should get C-like performance for simple calculations like this from numpy.

For example, your nested for loops can be replaced with a couple of numpy expressions...

I'm not terribly familiar with opencv, but I think the python bindings now have a numpy array interface, so your example above should be as simple as:

cv.PyrDown(img, dsimg)

data = np.asarray(dsimg)
blue, green, red = data.T

res = (green > (_RED_DIFF + red)) & (green > (_BLU_DIFF + blue))
res = res.astype(np.uint8) * 255

res = cv.fromarray(res)

(Completely untested, of course...) Again, I'm really not terribly familar with opencv, but nested python for loops are not the way to go about modifying an image element-wise, regardless...

Hope that helps a bit, anyway!

answered Aug 22, 2010 at 20:44

Joe Kington

287k73 gold badges621 silver badges474 bronze badges

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

5 Comments

bluth Over a year ago

This definitely speeds things up but I cannot seem to figure out why the result is more closely related to the threshold of my image and it does not find the green areas on the image as it seems it should?

Joe Kington Over a year ago

I may have misinterpreted what order the red, green, and blue channels of the image are in... (From your code above, it seemed like blue was the first channel and red was the last...) Maybe try unpacking the "data" array as red, green, blue = data.T? Just a wild guess... I may be completely wrong, there...

bluth Over a year ago

Yea, that was my initial thought, but changing the order doesn't produce a different result. The _RED_DIFF and _BLU are high values, 155, so there should not be many white pixels inside the image and yet there is; its almost all white when I put the _DIFF at 255? I can't seem to make sense of the results from res = (green > (_RED_DIFF + red)) & (green > (_BLU_DIFF + blue)).

Joe Kington Over a year ago

Hmm... It might be an overflow problem... In numpy, if you're dealing with uint8's, they overflow just like they would in C. If you're adding 155 to a uint8, there's a good chance that a lot of the values are overflowing to lower values. Maybe try casting things as uint16's to begin with? data = np.asarray(dsimg, dtype=np.uint16)?

bluth Over a year ago

I must say very clever. That did the trick. Thank you for all your help, very much appreciated.

Collectives™ on Stack Overflow

OpenCV Python binds incredibly slow iterations through image data

1 Answer 1

5 Comments

Your Answer

Linked

Hot Network Questions

Collectives™ on Stack Overflow

1 Answer 1

5 Comments

Your Answer

Sign up or log in

Post as a guest

Linked

Related