How to speed up a euclidean distance calculation between one pixel and many others

Question

Stephanie Diaz 2018-11-14

0
链接

此问题的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/429894-how-to-speed-up-a-euclidean-distance-calculation-between-one-pixel-and-many-others

评论： Guillaume 2018-11-14

I am calculating the euclidean distance between a pixel (x,y) and many other pixels in a matrix. I identify the x and y coordinates of the pixels beforehand, and save their x and y values in two column vectors x2 and y2, respectively I currently use the following:

distance = (sqrt((x2-x).^2 + (y2-y).^2))

This works, but is very slow. How can I speed this up?

1 个评论
显示 -1更早的评论隐藏 -1更早的评论

Guillaume 2018-11-14

在 MATLAB Online 中打开

This works, but is very slow.

How did you establish that it is slow?

>> coord = rand(1e6, 2);
>> tic; distance = sqrt((coord(:, 1) - coord(1, 1)).^2 + (coord(:, 2) - coord(1, 2)).^2); toc
Elapsed time is 0.029055 seconds.

As you can see it takes around 30 milliseconds on my computer to process a million coordinates.

Most likely, it's another part of your program that is slow, so show us your code.

Note that using hypot would be safer (albeit neglibly slower) than doing the calculation yourself:

distance = hypot(x2-x, y2-y);

请先登录，再进行评论。

请先登录，再回答此问题。