CODEWITHSUNDEEP
c++opencv
user14661890
user14661890

Reputation:

How to split an image into M x N tiles

How can I split a Mat object/image into M x N equal pieces and store the pieces from left to right and from top down, into a Mat vector using OpenCV and C++?

Many thanks

Upvotes: 2

Views: 1870

Answers (2)

Sdra
Sdra

Reputation: 2347

Inline with the solution given here for the opposite problem, this should do it:

#include <opencv2/opencv.hpp>

std::vector<cv::Mat> splitImage( cv::Mat & image, int M, int N )
{
  // All images should be the same size ...
  int width  = image.cols / M;
  int height = image.rows / N;
  // ... except for the Mth column and the Nth row
  int width_last_column = width  + ( image.cols % width  );
  int height_last_row   = height + ( image.rows % height );

  std::vector<cv::Mat> result;

  for( int i = 0; i < N; ++i )
  {
    for( int j = 0; j < M; ++j )
    {
      // Compute the region to crop from
      cv::Rect roi( width  * j,
                    height * i,
                    ( j == ( M - 1 ) ) ? width_last_column : width,
                    ( i == ( N - 1 ) ) ? height_last_row   : height );

      result.push_back( image( roi ) );
    }
  }

  return result;
}


int main()
{
  cv::Mat image = cv::imread( "image.png" );
  std::vector<cv::Mat> array_of_images = splitImage( image, 3, 2 );

  for( int i = 0; i < array_of_images.size(); ++i )
  {
    cv::imshow( "Image " + std::to_string( i ), array_of_images[i] );
  }
  cv::waitKey( 0 );
}

Upvotes: 1

SHR
SHR

Reputation: 8333

Here an example:

You take rectangles one by one in a loop and use crop to capture the selected part from the full picture.

Notice the full picture may not be divided perfectly, so one part in each row /column may be slightly bigger than the rest.

#include <opencv2/opencv.hpp>
#include <highgui.h>
#include <iostream>
#include <string>
#include <vector>

//usage: app pic.ext M N
int main(int argc, char* argv[])
{
    int M=0,N=0;
    
    M = std::stoi(argv[2]);
    N = std::stoi(argv[3]);
    
    cv::Mat img = cv::imread(argv[1]);
    if (img.empty())
    {
        std::cout << "failed to open the image" << std::endl;
        return -1;        
    }
    std::vector<std::vector<cv::Mat>> result;
    /* Regular size */
    int crop_width = img.size().width / M;
    int crop_height = img.size().height / N;
    /* column / row first part size when part size is not evenly divided*/
    int crop_width1 = img.size().width - (M-1)*crop_width;
    int crop_height1 = img.size().height - (N-1)*crop_height;
    
    int offset_x = 0;
    int offset_y = 0;
    for(int i = 0; i < M; i++) //rows
    {
        std::vector<cv::Mat> tmp;
        offset_x = 0;
        for(int j=0; j< N; j++) //columns
        {
            cv::Rect roi;
            roi.x = offset_x;
            roi.y = offset_y 
            roi.width = (j>0) ? crop_width :  crop_width1;
            roi.height = (i>0) ? crop_height : crop_height1;
            offset_x += roi.width;
            /* Crop the original image to the defined ROI */
            cv::Mat crop = img(roi);
            /* You can save part to a file: */
            //cv::imwrite(std::to_string(i) + "x" + std::to_string(j)+".png", crop);
            tmp.push_back(crop);
            
            
        }
        result.push_back(tmp);
        offset_y += (i>0) crop_height : crop_height1; 
    }
    return 0;
}

Upvotes: 0

Related Questions