nano.py

import cv2
import cv2.cv as cv
import numpy as np
import cmath

img = cv2.imread('nano.bmp', cv2.CV_LOAD_IMAGE_GRAYSCALE)

# Find particles
img = cv2.medianBlur(img, 3)
circles = cv2.HoughCircles(img, cv.CV_HOUGH_GRADIENT, 4, 10, param1=255, param2=20, minRadius=1, maxRadius=10)

# Round positions to integers
circles = np.uint16(np.around(circles))

# Draw circles
cimg = cv2.cvtColor(img, cv2.COLOR_GRAY2BGR)
for i in circles[0, :]:
    cv2.circle(cimg,(i[0],i[1]),i[2],(0,255,0),2)

# Find clusters
CLUSTER_DELTA = 25
clusters = []
count = []
added = 0
# Iterate through all circles
for i in circles[0, :]:
    added = 0
    k = 0
    # Check distance between a circle and cluster for every cluster, until:
    # a) all clusters are checked
    # b) some cluster is nearer than CLUSTER_DELTA
    for j in clusters[:]:
        print j
        x = j[0] - i[0]
        y = j[1] - i[1]
        d = cmath.sqrt(x*x + y*y).real
        if d < CLUSTER_DELTA:
            # 3. If some cluster is near enough, then:
            # a) we add circle to it
            # b) increase count (2D array in code) for it
            # c) we compute the cluster radius
            # d) we terminate looping through clusters
            j[0] = (j[0] + i[0])/2
            j[1] = (j[1] + i[1])/2
            j[2] = max(j[2] + i[2], d/2 + i[2]);
            count[k] += 1
            added = 1
            break
        k += 1
    if not added:
        # 4. If, after terminating looping through clusters,
        # no near cluster where discovered,
        # then this circle is a new cluster
        clusters.append(i)
        count.append(1)

# Draw clusters
k = 0
for i in clusters[:]:
    if count[k] > 1:
        cv2.circle(cimg,(i[0],i[1]),i[2],(255,0,0),2)
    k += 1



cv2.imshow('img', img)
cv2.imshow('cimg', cimg)
cv2.waitKey(0)
cv2.destroyAllWindows()