Multi-curve detection in noisy images in Python

Question

I have images similar to the one below. First of all, I am trying to detect the curves in these images. The curves I want to capture are marked on the image. Next, I want to fit these curves into the circle. I will use the radii of these circles as result. But I have problem with detecting curves in images. Any help much appreciated. Thanks in advance.

Input Image

Expected

Here's the code I'm using to detect and draw the curves:

import cv2
import numpy as np
from skimage.feature import peak_local_max
from skimage.morphology import watershed
from scipy import ndimage

image = cv2.imread("croppedImage.png")

img = cv2.medianBlur(image,13)

gray = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)

thresh = cv2.adaptiveThreshold(gray,255,cv2.ADAPTIVE_THRESH_MEAN_C,\
            cv2.THRESH_BINARY,45,0)

kernel = cv2.getStructuringElement(cv2.MORPH_ELLIPSE,(5,3))
kernel1 = np.ones((3, 3), np.uint8)
kernel2 = cv2.getStructuringElement(cv2.MORPH_ELLIPSE,(3,3))
dilate = cv2.dilate(thresh, kernel1, iterations=1)
erode = cv2.erode(dilate, kernel,iterations=1)

# Remove small noise by filtering using contour area
cnts = cv2.findContours(erode, cv2.RETR_TREE, cv2.CHAIN_APPROX_SIMPLE)
cnts = cnts[0] if len(cnts) == 2 else cnts[1]

for c in cnts:
    if cv2.contourArea(c) < 800:
        if len(c)>0:
            cv2.drawContours(thresh,[c], 0, (0,0,0), -1)
        
# Compute Euclidean distance from every binary pixel
# to the nearest zero pixel then find peaks
distance_map = ndimage.distance_transform_edt(erode)
local_max = peak_local_max(distance_map, indices=False, min_distance=1, labels=thresh)

# Perform connected component analysis then apply Watershed
markers = ndimage.label(local_max, structure=np.ones((3, 3)))[0]
labels = watershed(-distance_map, markers, mask=erode)

# Iterate through unique labels
for label in np.unique(labels):
    if label == 0:
        continue

    # Create a mask
    mask = np.zeros(thresh.shape, dtype="uint8")
    mask[labels == label] = 255

    # Find contours and determine contour area
    cnts = cv2.findContours(mask.copy(), cv2.RETR_EXTERNAL, cv2.CHAIN_APPROX_SIMPLE)
    cnts = cnts[0] if len(cnts) == 2 else cnts[1]
    c = max(cnts, key=cv2.contourArea)
    
    cv2.drawContours(image, [c], -1, (36,255,12), -1)

        

cv2.imwrite('Results/drawedImage.png',image)

thresh = 155
im_bw = cv2.threshold(image, thresh, 255, cv2.THRESH_BINARY)[1]

cv2.imwrite("Results/binary.png",im_bw)

Result Image

Binary Result

From the images like below, I can fit circles. But I don't have clean images like this one.

gray_blurred = cv2.GaussianBlur(img,(11,11),0)

ret3,thresh= cv2.threshold(gray_blurred,100,255,cv2.THRESH_BINARY+cv2.THRESH_OTSU)

# Apply Hough transform on the blurred image. 
detected_circles = cv2.HoughCircles(thresh,  
                   cv2.HOUGH_GRADIENT, 1, 80, param1 = 20, 
               param2 = 9, minRadius = 120, maxRadius = 200) 

  
# Draw circles that are detected. 
if detected_circles is not None: 
  
    # Convert the circle parameters a, b and r to integers. 
    detected_circles = np.uint16(np.around(detected_circles)) 

    for pt in detected_circles[0, :]: 
        a, b, r = pt[0], pt[1], pt[2] 
        
        # Draw the circumference of the circle. 
        cv2.circle(img, (a, b), r, (0, 255, 0), 2) 
  
        # Draw a small circle (of radius 1) to show the center. 
        cv2.circle(img, (a, b), 1, (0, 0, 255), 3) 
 
else:
    print("Circle is not found")