c++ - Opencv: Edge detection, Dilation and Mass ceter drawing -

my work based on images array of dots (fig. 1), , final result shown in fig. 4. explain work step step.

fig. 1 original image

step 1: detect edge of every object, including dots , "ring" want delete better performance. , result of edge detection shown in fig.2. used canny edge detector didn't work light-gray dots. my first question how close contours of dots , reduce other noise as possible?

fig. 2 edge detection

step 2: dilate every object. didn't find way fill holes, dilate them directly. shown in fig.3, holes seem enlarged , other noise. my second question how fill or dilate holes in order make them filled circles in same/similar size?

fig. 3 dilation

step 3: find , draw mass center of every dot. shown in fig. 4, due coarse image processing, there exist mark of "ring" , of dots shown in 2 white pixels. the result wanted should show dots , 1 white pixel 1 dot.

fig. 4: mass centers

here code these 3 steps. can make work better?

#include "opencv2/imgproc/imgproc.hpp" #include "opencv2/highgui/highgui.hpp" #include <stdlib.h> #include <stdio.h> #include <cv.h> #include <highgui.h> using namespace std; using namespace cv;  // global variables mat src, edge, dilation; int dilation_size = 2;  // function header void thresh_callback(int, void*);  int main(int argc, char* argv) {     iplimage* img = cvloadimage("c:\\dot1.bmp", 0);         // dot1.bmp = fig. 1      // perform canny edge detection     cvcanny(img, img, 33, 100, 3);      // iplimage mat     mat imgmat(img);     src = img;      namedwindow("step 1: edge", cv_window_autosize);     imshow("step 1: edge", src);      // apply dilation operation     mat element = getstructuringelement(2, size(2 * dilation_size + 1, 2 * dilation_size + 1),                    point(dilation_size, dilation_size));     // dilation_type = morph_ellipse     dilate(src, dilation, element);     // imwrite("c:\\dot1_dilate.bmp", dilation);                  namedwindow("step 2: dilation", cv_window_autosize);     imshow("step 2: dilation", dilation);      thresh_callback( 0, 0 );      waitkey(0);     return 0; }  /* function thresh_callback */ void thresh_callback(int, void*) {     vector<vector<point>> contours;     vector<vec4i> hierarchy;      // find contours     findcontours(dilation, contours, hierarchy, cv_retr_tree, cv_chain_approx_simple, point(0, 0));      // moments     vector<moments> mu(contours.size());     for(int = 0; < contours.size(); i++) {         mu[i] = moments(contours[i], false);     }      // mass centers     vector<point2f> mc(contours.size());     for(int = 0; < contours.size(); i++) {         mc[i] = point2f(mu[i].m10/mu[i].m00 , mu[i].m01/mu[i].m00);      }      // draw mass centers     mat drawing = mat::zeros(dilation.size(), cv_8uc1);     for( int = 0; i< contours.size(); i++ ) {         scalar color = scalar(255, 255, 255);         line(drawing, mc[i], mc[i], color, 1, 8, 0);     }      namedwindow("step 3: mass centers", cv_window_autosize);     imshow("step 3: mass centers", drawing); } 

there few things can improve results. reduce noise in image, can apply median blur before applying canny operator. common de-noising technique. also, try avoid using c api , iplimage.

    cv::mat img = cv::imread("c:\\dot1.bmp", 0);         // dot1.bmp = fig. 1      cv::medianblur(img, img, 7);      // perform canny edge detection     cv::canny(img, img, 33, 100); 

this reduces amount of noise in edge image:

to better retain original sizes of dots, can perform few iterations of morphological closing smaller kernel rather dilation. reduce joining of dots circle:

// replaces call dilate() cv::morphologyex(src, dilation, morph_close, cv::noarray(),cv::point(-1,-1),2); 

this perform 2 iterations 3x3 kernel, indicated using cv::noarray().

the result cleaner, , dots filled:

leaving rest of pipeline unmodified gives final result. there still few spurious mass centers circle, considerably fewer original method:

if wanted attempt removing circle results entirely, try using cv::houghcircles() , adjusting parameters until result. might have difficulties because entire circle not visible in image, segments, recommend experiment it. if did detect innermost circle, use mask filter out external mass centers.