visual-c++ - 使用单个模板匹配多个对象

Question

如何使用单个模板匹配多个对象？我正在尝试使用树的中心作为模板来匹配多棵香蕉树。我的程序只匹配一次，我希望匹配航拍图像中香蕉树的所有出现。`

#include "opencv2/highgui/highgui.hpp"
#include "opencv2/imgproc/imgproc.hpp"
#include <iostream>
#include <stdio.h>

using namespace std;
using namespace cv;

/// Global Variables
Mat img; Mat templ; Mat result;
const char* image_window = "Source Image";
const char* result_window = "Result window";

int match_method;
int max_Trackbar = 5;

/// Function Headers
void MatchingMethod( int, void* );

/**
 * @function main
 */
int main( int, char** argv )
{
  /// Load image and template
  img = imread( argv[1], 1 );
  templ = imread( argv[2], 1 );

  /// Create windows
  namedWindow( image_window, WINDOW_AUTOSIZE );
  namedWindow( result_window, WINDOW_AUTOSIZE );

  /// Create Trackbar
  const char* trackbar_label = "Method: \n 0: SQDIFF \n 1: SQDIFF NORMED \n 2: TM CCORR \n 3: TM CCORR NORMED \n 4: TM COEFF \n 5: TM COEFF NORMED";
  createTrackbar( trackbar_label, image_window, &match_method, max_Trackbar, MatchingMethod );

  MatchingMethod( 0, 0 );

  waitKey(0);
  return 0;
}

/**
 * @function MatchingMethod
 * @brief Trackbar callback
 */
void MatchingMethod( int, void* )
{
  /// Source image to display
  Mat img_display;
  img.copyTo( img_display );

  /// Create the result matrix
  int result_cols =  img.cols - templ.cols + 1;
  int result_rows = img.rows - templ.rows + 1;

  result.create( result_cols, result_rows, CV_32FC1 );

  /// Do the Matching and Normalize
  matchTemplate( img, templ, result, match_method );
  normalize( result, result, 0, 1, NORM_MINMAX, -1, Mat() );

  /// Localizing the best match with minMaxLoc
  double minVal; double maxVal; Point minLoc; Point maxLoc;
  Point matchLoc;

  minMaxLoc( result, &minVal, &maxVal, &minLoc, &maxLoc, Mat() );


  /// For SQDIFF and SQDIFF_NORMED, the best matches are lower values. For all the other methods, the higher the better
  if( match_method  == TM_SQDIFF || match_method == TM_SQDIFF_NORMED )
    { matchLoc = minLoc; }
  else
    { matchLoc = maxLoc; }

  /// Show me what you got
  rectangle( img_display, matchLoc, Point( matchLoc.x + templ.cols , matchLoc.y + templ.rows ), Scalar::all(0), 2, 8, 0 );
  rectangle( result, matchLoc, Point( matchLoc.x + templ.cols , matchLoc.y + templ.rows ), Scalar::all(0), 2, 8, 0 );

  imshow( image_window, img_display );
  imshow( result_window, result );

  return;
}

`

Answer 1

在 Saikat(和 Bartlett)的代码中使用如下行

result.at<float>(minLoc.x,minLoc.y)=1.0;

和类似的行有下一个缺点:代码屏蔽了唯一的极值像素，然后进行下一个循环可能会找到相同的对象，将一个像素移到一边。我建议用模板大小的矩形来掩盖结果。此代码启用控制相邻物体的重叠程度。

void matchingMethod(Mat& img,  const Mat& templ,  int     match_method)
{
    /// Source image to display
    Mat img_display; Mat result;
   if(img.channels()==3)
        cvtColor(img, img, cv::COLOR_BGR2GRAY);
    img.copyTo( img_display );//for later show off

    /// Create the result matrix - shows template responces
    int result_cols = img.cols - templ.cols + 1;
    int result_rows = img.rows - templ.rows + 1;
    result.create( result_cols, result_rows, CV_32FC1 );

    /// Do the Matching and Normalize
    matchTemplate( img, templ, result, match_method );
    normalize( result, result, 0, 1, NORM_MINMAX, -1, Mat() );

    /// Localizing the best match with minMaxLoc
    double minVal; double maxVal; 
    Point minLoc; Point maxLoc;
    Point matchLoc;


    //in my variant we create general initially positive mask 
    Mat general_mask=Mat::ones(result.rows,result.cols,CV_8UC1);

    for(int k=0;k<5;++k)// look for N=5 objects
    {
        minMaxLoc( result, &minVal, &maxVal, &minLoc, &maxLoc, general_mask); 
        //just to visually observe centering I stay this part of code:
        result.at<float>(minLoc ) =1.0;//
        result.at<float>(maxLoc ) =0.0;//

        // For SQDIFF and SQDIFF_NORMED, the best matches are lower values. 
         //For all the other methods, the higher the better
        if( match_method  == CV_TM_SQDIFF || match_method ==     CV_TM_SQDIFF_NORMED )
            matchLoc = minLoc;
        else
            matchLoc = maxLoc;
                                //koeffitient to control neiboring:
        //k_overlapping=1.- two neiboring selections can overlap half-body of     template
        //k_overlapping=2.- no overlapping,only border touching possible
        //k_overlapping>2.- distancing
        //0.< k_overlapping <1.-  selections can overlap more then half 
        float k_overlapping=1.7f;//little overlapping is good for my task

        //create template size for masking objects, which have been found,
        //to be excluded in the next loop run
        int template_w= ceil(k_overlapping*templ.cols);
        int template_h= ceil(k_overlapping*templ.rows);
        int x=matchLoc.x-template_w/2;
        int y=matchLoc.y-template_h/2;

        //shrink template-mask size to avoid boundary violation
        if(y<0) y=0;
        if(x<0) x=0;
        //will template come beyond the mask?:if yes-cut off margin; 
        if(template_w + x  > general_mask.cols) 
            template_w= general_mask.cols-x;
        if(template_h + y  > general_mask.rows) 
            template_h= general_mask.rows-y;

                               //set the negative mask to prevent repeating
        Mat template_mask=Mat::zeros(template_h,template_w, CV_8UC1);
        template_mask.copyTo(general_mask(cv::Rect(x, y, template_w, template_h)));

        /// Show me what you got on main image and on result (
        rectangle( img_display,matchLoc , Point( matchLoc.x + templ.cols ,    matchLoc.y + templ.rows ), Scalar::all(0), 2, 8, 0 );
        //small correction here-size of "result" is smaller
        rectangle( result,Point(matchLoc.x- templ.cols/2,matchLoc.y-     templ.rows/2) , Point( matchLoc.x + templ.cols/2 , matchLoc.y + templ.rows/2 ),     Scalar::all(0), 2, 8, 0 );
    }//for k= 0--5 
}