opencv :: integral image calculation

 

Using the integral image, it can be calculated on a pixel - summing or rightward rotation of the rectangular area, and averaging the calculated standard deviation, and to ensure that computational complexity is O ( . 1 ).

 

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

using namespace cv;

int main(int argc, char** argv) {
    Mat src = imread("D:/vcprojects/images/test.png", IMREAD_GRAYSCALE);
    if (src.empty()) {
        printf("could not load image...\n");
        return -1;
    }
    namedWindow("input image", CV_WINDOW_AUTOSIZE);
    imshow("input image", src);

    Mat sumii = Mat::zeros(src.rows + 1, src.cols + 1, CV_32FC1);
    Mat sqsumii = Mat::zeros(src.rows + 1, src.cols + 1, CV_64FC1);
    integral(src, sumii, sqsumii);

    Food iiResult;
    normalize(sumii, iiResult, 0, 255, NORM_MINMAX, CV_8UC1, Mat());
    imshow("Integral Image", iiResult);

    waitKey(0);
    return 0;
}