Ask Your Question

Extract blobs that intersect between 2 images

asked 2018-06-25 07:49:06 -0500

sazr gravatar image

In OpenCV, how would I be able to extact all the blobs in image A that collide with the blobs/markers in image B? Is there an easy way? One way I can think of is to use Marker Controlled Watershed Segmentation. I haven't tried/done that before - do I need to label/categorise each marker in image b first? Is there a simpler (and more efficient way) to extract my 'intersection' blobs? Ie, is there an easier way to achieve the 3rd image?

Image A:

image description

Image B:

image description

My desired result is below:

image description

edit retag flag offensive close merge delete

1 answer

Sort by ยป oldest newest most voted

answered 2018-06-25 11:56:15 -0500

sjhalayka gravatar image

updated 2018-06-25 18:27:55 -0500

I call this algorithm the poor man's segmentation. It should be made known that there is noise (JPEG artifacts) in the input images, but the code overcomes that, and works as requested. In the future, please save some time and grief by using PNG files! If you find this answer to be useful, please mark it as correct and upvote it. Thank you. :)

The output is:

image description

The code is given below. In the function call threshold(frame_a, frame_a,127, 255, THRESH_BINARY);, the 127 value can be altered to suit one's taste:

#pragma comment(lib, "opencv_world340.lib")

#include <opencv2/opencv.hpp>
using namespace cv;

#include <iostream>
using namespace std;

int main(void)
    Mat frame_a = imread("frame_a.jpg", CV_LOAD_IMAGE_GRAYSCALE);
    Mat frame_b = imread("frame_b.jpg", CV_LOAD_IMAGE_GRAYSCALE);

    if (frame_a.empty() || frame_b.empty())
        cout << "Could not read input files" << endl;
        return 1;

    // get rid of JPEG artifacts.
    // please use PNG next time!!!
    threshold(frame_a, frame_a, 127, 255, THRESH_BINARY);

    int fill_colour = 19;

    for (int j = 0; j < frame_b.rows; j++)
        for (int i = 0; i < frame_b.cols; i++)
            unsigned char colour =<unsigned char>(j, i);

            if (colour == 255)
                floodFill(frame_a, Point(i, j), Scalar(fill_colour));

    Mat frame_out(frame_a.rows, frame_a.cols, CV_8UC1);

    for (int j = 0; j < frame_a.rows; j++)
        for (int i = 0; i < frame_a.cols; i++)
            unsigned char colour =<unsigned char>(j, i);

            if (colour == fill_colour)
      <unsigned char>(j, i) = 255;
      <unsigned char>(j, i) = 0;

    //imshow("frame a", frame_a);
    //imshow("frame b", frame_b);
    imshow("frame out", frame_out);

    return 0;
edit flag offensive delete link more
Login/Signup to Answer

Question Tools

1 follower


Asked: 2018-06-25 07:49:06 -0500

Seen: 63 times

Last updated: Jun 25 '18