OPENCV ::查找两个同心圆之间的所有点

Question

I have two concentric circles a and b, I detected the circles from the image, I know their Radii. What I want is the Points between the two concentric cirlces, So that I can extract the RGB values from those Points. Any help would be helpfull.

Answer 1

The indices ( x, y ) of the points between the two concentric circles (center ( x_0, y_0 ), radii r_0, r_1 with r_1 > r_0 ) have to fulfill the following condition:

(x-x_0) * (x-x_0) + (y-y_0) * (y-y_0) >= r_0 * r_0
(x-x_0) * (x-x_0) + (y-y_0) * (y-y_0) <= r_1 * r_1

So when looping over all point of an image you can figure out which points to process:

for (int x=0; x<img.rows; x++)
{
   for (int y=0; y<img.cols; y++)
   {
       double dd = (x-x_0) * (x-x_0) + (y-y_0) * (y-y_0);
       if (dd < r_0 * r_0 || dd > r_1 * r_1)
         continue;

       // Do what you have to do with the points between the two circles
   }
}

Answer 2

Pseudocode:

Make a Mat of the same size as your source image. 
Draw the two filled circles with the built-in OpenCV circle drawing routine. 
  (large circle in white, then the small in black) 
Use that as a mask, that you multiply on your source image. 
(Depending on which pixels you wish to keep, the mask may have to be inverted.)

Answer 3

If you need to visit all pixels and somehow gather their colour values, and you only want to visit the pixels in question as speed is crucial, I would suggest that you look into the midpoint circle algorithm authored by Bresenham . You could specialize this so that you only get 2 or 4 points per pixel line, and traverse each line's pair(s) of points. Should be possible O(n), where n is height in pixels (diameter, ie largest radius), so you'd look at o(m), m being the number of pixels between the two pixels, instead of n^2. Would matter a lot if you have large images and radii close to each other.