Unable to find correct circles with cv2.HoughCircles

Question

When I run the code below, I am not able to find consistent circles in the image. The image I am using as input is:

import numpy as np
import matplotlib.pyplot as plt
import cv2
img = cv2.imread("pipe.jpg")


# convert the image to RGB
img = cv2.cvtColor(img, cv2.COLOR_BGR2RGB)
# copy the RGB image
cimg = img.copy()
# convert the RGB image to grayscale
img = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)
# detect circles using hough transformation
circles = cv2.HoughCircles(image=img, method=cv2.HOUGH_GRADIENT, dp=3,
                        minDist=60, param1=100, param2=39, maxRadius=200)

for co,i in enumerate(circles[0, :],start=1):
    i = [round(num) for num in i]
    cv2.circle(cimg,(i[0],i[1]),i[2],(0,255,0),2)
    cv2.circle(cimg,(i[0],i[1]),2,(0,0,255),3)
print("Number of circles detected:", co)
plt.imshow(cimg)
plt.show()

The result I get is:

Answer 1

As a pre-processing step, you usually smooth the image prior to the detection. Not smoothing will prompt a lot of false detections. Sometimes you don't see pre-smoothing in some tutorials because the images dealt with have nice clean edges with very little noise. Try using a median blur or Gaussian blur before you perform the detection.

Therefore, try something like:

import cv2

img = cv2.imread("pipe.jpg")

# convert the image to RGB
img = cv2.cvtColor(img, cv2.COLOR_BGR2RGB)
# copy the RGB image
cimg = img.copy()
# convert the RGB image to grayscale
img = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)

### NEW - Smooth the image first
blur = cv2.GaussianBlur(img, (5, 5), 0)
# or try
# blur = cv2.medianBlur(img, 5)

# detect circles using hough transformation
circles = cv2.HoughCircles(image=blur, method=cv2.HOUGH_GRADIENT, dp=3,
                        minDist=60, param1=100, param2=39, maxRadius=200)

Other than that, getting the detection to find all of the circles in the image is subject to playing with the hyperparameters of the Circular Hough Transform and through trial and error.