【深度学习】基于区域生长的图像分割算法！

#include #include 
using namespace std;using namespace cv;// 定义区域生长分割算法bool RegionGrowing(Mat img, Mat& result, Point2i seed, int threshold) {  // 将图像全部设置为黑    result = Mat::zeros(img.size(), CV_8UC1);     // 设置种子点范围条件    if (seed.x < 0 || seed.y < 0 || seed.y > img.rows - 1 || seed.x > img.cols - 1) {        return false;    }    // 种子点集    vector seeds;     // 压入初始种子点    seeds.push_back(seed);     // 种子点设置为白    result.ptr(seed.y)[seed.x] = 255;     // 8连接生长方向    int growDirections[8][2] = { {-1,-1}, {0,-1}, {1,-1}, {-1,0}, {1,0}, {-1,1}, {0,1}, {1,1} }; 
    // 开始生长    while (!seeds.empty()) {        // 取出一个种子点作为现在循环的初始种子点        Point2i seed_current = seeds.back();        seeds.pop_back();
        // 遍历各生长方向的邻点        for (int i = 0; i < 8; i++) {            Point2i neighborPoint = { seed_current.x + growDirections[i][0], seed_current.y + growDirections[i][1] }; // 邻点            if (neighborPoint.x < 0 || neighborPoint.y < 0 || neighborPoint.x > img.cols - 1 || neighborPoint.y > img.rows - 1) { // 邻点超出范围                continue;            }            if ((result.ptr(neighborPoint.y)[neighborPoint.x] == 0) && abs(img.ptr(neighborPoint.y)[neighborPoint.x] - img.ptr(seed.y)[seed.x]) < threshold) {                // 设置为种子点                result.ptr(neighborPoint.y)[neighborPoint.x] = 255; // 设置为白色                seeds.push_back(neighborPoint); // 压入种子集            }        }    }
    return true;}