gaussian_filter_by_oepncv.cpp

#include <iostream>
#include <math.h>
#include <opencv2/opencv.hpp>

using namespace cv;

Mat sourceImg;
Mat resultImg;

//가우시안 함수
double gaussian_func(int y, int x, double sd){
    double sd2 = pow(sd,2);
    return exp(-(x*x+y*y)/(2*sd2))/(asin(1)*2*2*sd2);
}

unsigned char gaussian_filter(int y, int x, double sd){
    int w = ceil(sd*6);
    
    if(w %2 == 0)
        w++;
    
    int sum = 0;
    for(int ny = -w/2; ny <= w/2; ny++){
        for(int nx = -w/2 ; nx <= w/2; nx++){
            int ty = y+ny;
            int tx = x+nx;
            
             //ty와 tx가 범위 밖을 나갔을 때 테두리 선에서 선대칭 되는 위치에 있는 픽셀값을 이용했습니다.
            if(ty<0)
                ty *= -1;
            if(ty>=sourceImg.rows)
                ty = 2*sourceImg.rows-ty;
            if(tx<0)
                tx *= -1;
            if(tx>=sourceImg.cols)
                tx = 2*sourceImg.cols-tx;
            
            sum += (int)sourceImg.at<unsigned char>(ty,tx)*gaussian_func(ny, nx, sd);
        }
    }
    return sum;
}

int main(){
    double sd;
    scanf("%lf", &sd);
    
    sourceImg = imread("/Users/clsrn1581/Desktop/noise.png", 0);
    resultImg = Mat(sourceImg.rows, sourceImg.cols, CV_8UC1, 1);
    
    
    for(int y = 0 ;y < sourceImg.rows; y++){
        for(int x = 0; x < sourceImg.cols; x++){
            resultImg.at<unsigned char>(y,x) = gaussian_filter(y,x, sd);
        }
    }
    
    namedWindow("Source Image");
    namedWindow("Result Image");
    
    imshow("Source Image", sourceImg);
    imshow("Result Image", resultImg);
    
    int key = waitKey();
    
    destroyAllWindows();
    
    return 0;
}