OpenCV circle like shape detection and its area - c++

Question

I have an image with one circle like shape that contains another similar shape. I am trying find the areas of those two shapes. I am using openCv c++ Hough circle detection, but it does not detect the shapes. Is there any other functions in OpenCV can be used to detect the shapes and find the ares?
[EDIT] The image has been added.
Here is my sample code
int main()
{
Mat src, gray;
src = imread( "detect_circles_simple.jpg", 1 );resize(src,src,Size(640,480));
cvtColor( src, gray, CV_BGR2GRAY );
// Reduce the noise so we avoid false circle detection
GaussianBlur( gray, gray, Size(9, 9), 2, 2 );
vector<Vec3f> circles;
// Apply the Hough Transform to find the circles
HoughCircles( gray, circles, CV_HOUGH_GRADIENT, 1, 30, 200, 50, 0, 0 );
cout << "No. of circles : " << circles.size()<<endl;
// Draw the circles detected
for( size_t i = 0; i < circles.size(); i++ )
{
Point center(cvRound(circles[i][0]), cvRound(circles[i][1]));
int radius = cvRound(circles[i][2]);
circle( src, center, 3, Scalar(0,255,0), -1, 8, 0 );// circle center
circle( src, center, radius, Scalar(0,0,255), 3, 8, 0 );// circle outline
cout << "center : " << center << "\nradius : " << radius << endl;
}
exit(0);
// Show your results
namedWindow( "Hough Circle Transform Demo", CV_WINDOW_AUTOSIZE );
imshow( "Hough Circle Transform Demo", src );
waitKey(0);
return 0;
}

Answer 1

I have a similar approach.
img1 = cv2.imread('disc1.jpg', 1)
img2 = img1.copy()
img = cv2.cvtColor(img1,cv2.COLOR_BGR2GRAY)
#--- Blur the gray scale image
img = cv2.GaussianBlur(img,(5, 5),0)
#--- Perform Canny edge detection (in my case lower = 84 and upper = 255, because I resized the image, may vary in your case)
edges = cv2.Canny(img, lower, upper)
cv2.imshow('Edges', edges )
#---Find and draw all existing contours
_, contours , _= cv2.findContours(edges, cv2.RETR_TREE, 1)
rep = cv2.drawContours(img1, contours, -1, (0,255,0), 3)
cv2.imshow(Contours',rep)
Since you are analyzing the shape of a circular edge, determining the eccentricity of your contours will help in this case.
#---Determine eccentricity
cnt = contours
for i in range(0, len(cnt)):
ellipse = cv2.fitEllipse(cnt[i])
(center,axes,orientation) =ellipse
majoraxis_length = max(axes)
minoraxis_length = min(axes)
eccentricity=(np.sqrt(1-(minoraxis_length/majoraxis_length)**2))
cv2.ellipse(img2,ellipse,(0,0,255),2)
cv2.imshow('Detected ellipse', img2)
Now based on the value given by the eccentricity variable you can come to a conclusion whether your contour is circular or not. The threshold depends on what you consider to be circular or an approximate circle.

Answer 2

If you have complete shapes (the edge completely or very nearly joins) it is generally easier to edge detect -> contour -> analyse the contour shape.
Hough lines or circles are very useful when you only have small fragments of a line or circle, but can be tricky to tune
edit: Try cv::adaptiveThreshold to get the edges, then cv::findContours.
For each contour compare the area to the perimeter to see if it is the right size to be your target. Then do cv::fitEllipse to check if it is a circle and get the accurate center. FindCOntours also has a mode which tells you which contours are inside which others, so you can easily find one circle inside another.
You might (depending on lighting) find the same circle with 2 or more contours, ie. for the inner and outer edge.