Particle analyze in binary image

edit

As neat as contours are, you probably want to use connectedComponentsWithStats

You pass it the binary image and it returns the number of white spots. The stats it returns include width, height, X and Y centroid, and area. In other words, exactly what you want, in one function call.

Here is a C++ and Python code to get the contours, particle count, particle area, and particle centres. It's an implementation of procton's algorithm, so make sure to mark their answer as correct. Let me know if you're a Python coder instead.

#include <opencv2/opencv.hpp>
using namespace cv;
#pragma comment(lib, "opencv_world331.lib")

#include <iostream>
#include <vector>
using namespace std;



int main(void)
{
    Mat frame = imread("particles.png", CV_LOAD_IMAGE_GRAYSCALE);

    if (frame.empty())
    {
        cout << "Error loading image file" << endl;
        return -1;
    }

    threshold(frame, frame, 127, 255, THRESH_BINARY);

    vector<vector<Point> > contours;
    vector<Vec4i> hierarchy;

    findContours(frame, contours, hierarchy, CV_RETR_TREE, CV_CHAIN_APPROX_SIMPLE, Point(0, 0));

    vector<double> areas(contours.size());

    for (int i = 0; i < contours.size(); i++)
        areas[i] = contourArea(contours[i]);

    vector<Point2d> mass_centres(contours.size());

    for (int i = 0; i < contours.size(); i++)
    {
        const Moments mu = moments(contours[i], false);
        mass_centres[i] = Point2d(mu.m10 / mu.m00, mu.m01 / mu.m00);
    }

    cout << "Num particles: " << contours.size() << endl;

    for (int i = 0; i < contours.size(); i++)
        cout << "area " << (i + 1) << ": " << areas[i] << endl;

    for (int i = 0; i < contours.size(); i++)
        cout << "centre " << (i + 1) << ": " << mass_centres[i].x << " " << mass_centres[i].y << endl;

    return 0;
}

Here is the Python code:

import cv2
import numpy

frame = cv2.imread('blobs.png')

if frame is None:
    print('Error loading image')
    exit()

frame = cv2.cvtColor(frame, cv2.COLOR_BGR2GRAY)
ret, frame = cv2.threshold(frame, 127, 255, cv2.THRESH_BINARY)

frame, contours, hierarchy = cv2.findContours(frame, cv2.RETR_TREE, cv2.CHAIN_APPROX_SIMPLE)

areas = []

for i in range(0, len(contours)):
    areas.append(cv2.contourArea(contours[i]))

mass_centres_x = []
mass_centres_y = []

for i in range(0, len(contours)):
    M = cv2.moments(contours[i], 0)
    mass_centres_x.append(int(M['m10']/M['m00']))
    mass_centres_y.append(int(M['m01']/M['m00']))

print 'Num particles: ', len(contours)

for i in range(0, len(contours)):
    print 'Area', (i + 1), ':', areas[i]

for i in range(0, len(contours)):
    print 'Centre',(i + 1),':', mass_centres_x[i], mass_centres_y[i]    

cv2.imshow("Frame", frame)

cv2.waitKey(0)

You can do the following:

1. find contours on this image, which is already binarized with black background so it's ready for contouring (https://docs.opencv.org/3.3.1/d4/d73/...)

2. for each contour there are APIs to calculate perimeter, area and center of mass (moments) https://docs.opencv.org/3.3.1/dd/d49/...

3. Of course you will easily be able to count contours